The Computer Incident Response Center Luxembourg (CIRCL) is a government-driven initiative designed to provide a systematic response facility to computer security threats and incidents.
- Website
- https://www.circl.lu
- info@circl.lu
- GNA identifier
- GNA-1 GCVE registry Recent publications
Recent vulnerabilities
355 GCVE records assigned by this organization as GNA-1GCVE-1-2026-20356
Vulnerability from gna-1 – Published: 2026-11-17 16:10 – Updated: 2026-11-17 16:10| URL | Tags |
|---|---|
| https://github.com/kool277/nginx/commit/7b185de7a… | patch |
| Vendor | Product | Version | |
|---|---|---|---|
| nginx | nginx HTTP/2 module |
Affected:
unspecified
cpe:2.3:a:nginx:nginx_http\/2_module:*:*:*:*:*:*:*:* |
qwen3.8:27b
2 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-17 16:08 - Model
qwen3.8:27b- Input
-
https://github.com/kool277/nginx/commit/7b185de7a3f15e545888380cae25b81594f490e1.patch
718197a690d2… - Confidence
- medium
| Commit | Subject | Patch SHA-256 |
|---|---|---|
7b185de7a3f1
|
HTTP/2: fixed skipped header fields after a stream reset. | 718197a690d2… |
Fix summary
The fix corrects the HTTP/2 state machine so that when a stream is reset mid-header-block, subsequent literal fields are properly marked as skipped (state.field_start set to NULL) and their true encoded length is recorded in state.header. The header processing function now only computes name/value lengths from field pointers for fields that were actually read, preventing both the false zero-length rejection that caused connection-level GOAWAY and the incorrect header list size accounting for skipped fields.
Patch summary
In ngx_http_v2_state_field_len(), when the stream is NULL and the field is non-indexed (i.e., being skipped), the patch sets state.field_start to NULL and records the field's encoded length into state.header.name.len or state.header.value.len (with data set to NULL) before calling the skip handler. In ngx_http_v2_state_process_header(), the name and value length assignments from field_start/field_end are now guarded by a NULL check on state.field_start, so skipped fields retain the length recorded during the skip phase. A new regression test (tests/regression/h2_reset_stream_skipped_field.t) verifies that a stream reset followed by a literal field does not produce GOAWAY, that the connection remains usable, and that a skipped field with an empty name still triggers PROTOCOL_ERROR.
CVSS rationale
AV:N: The vulnerability is exploitable over the network via a standard HTTP/2 connection. AC:L: The attack requires only sending a single crafted HEADERS frame with an invalid pseudo-header followed by a literal field; no race conditions or special timing are needed. AT:N: No manipulation of the attack target is required. PR:N: No authentication is needed to send HTTP/2 requests to a server. UI:N: No user interaction is required. VC:N, VI:N: No confidentiality or integrity impact is evident from the patch. VA:L: The impact is the termination of a single HTTP/2 connection (GOAWAY), which is a limited availability impact per instance; an attacker could repeat the attack across multiple connections but each instance only loses one connection. SC:N, SI:N, SA:N: No secondary component impact is indicated.
Weakness rationale
- CWE-20 The HTTP/2 state machine does not properly handle the transition from active field parsing to field skipping when a stream is reset mid-header-block. The stale pointer state leads to incorrect validation decisions (false zero-length rejection) and incorrect size accounting, both of which are consequences of failing to properly validate and track the state of skipped input.
- CWE-665 When entering the field-skip path, state.field_start and state.field_end are not re-initialized to reflect the skipped state. They retain values from a prior field, causing downstream logic to operate on stale data. The fix explicitly sets field_start to NULL to mark the field as skipped.
Attack pattern rationale
- CAPEC-107 The attacker sends a crafted HTTP/2 header block that triggers a stream reset (via an invalid pseudo-header) followed by a literal field, manipulating the server's internal HTTP/2 state machine into an inconsistent state where stale pointers cause a false protocol error and connection teardown. This is the closest CAPEC to the observed attack: the attacker exploits the application's state management logic by forcing a specific state transition sequence. Uncertainty: no CAPEC specifically covers HTTP/2 protocol state manipulation; CAPEC-107 is the best available match for exploiting state machine inconsistencies.
Assumptions to verify
- The affected nginx version range is not specified in the patch; the commit references nginx/nginx#1977 but no version boundary is provided.
- The CVSS VA:L rating assumes the impact is limited to a single HTTP/2 connection per attack instance; if the server has a very small connection pool, the practical DoS impact could be higher.
- CAPEC-107 (State Manipulation) is the closest available CAPEC; no CAPEC specifically addresses HTTP/2 protocol state machine manipulation, so this mapping is approximate.
- The secondary impact of incorrect header list size accounting (skipped fields charged with wrong lengths) is assessed as not independently exploitable for a distinct security impact beyond the primary DoS vector.
- The patch is from a fork (kool277/nginx); it is assumed to correspond to an upstream nginx fix, but the exact upstream commit and version are not confirmed in the supplied metadata.
- No evidence of authentication requirements, deployment topology, or multi-tenant context is present; the analysis assumes a standard single-tenant nginx HTTP/2 server.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
6 | 11 | medium | 6 |
- 2026-11-14 11:00 UTC Fix developed Corrective change authored (7b185de7a3f15e545888380cae25b81594f490e1): HTTP/2: fixed skipped header fields after a stream reset. https://github.com/kool277/nginx/commit/7b185de7a3f15e545888380cae25b81594f490e1.patch
- 2026-11-17 16:07 UTC Fix available Patch retrieved from the supplied URL; retrieval time is used as the default availability time. https://github.com/kool277/nginx/commit/7b185de7a3f15e545888380cae25b81594f490e1.patch
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:nginx:nginx_http\\/2_module:*:*:*:*:*:*:*:*"
],
"modules": [
"src/http/v2/ngx_http_v2.c"
],
"product": "nginx HTTP/2 module",
"programFiles": [
"src/http/v2/ngx_http_v2.c"
],
"repo": "https://github.com/kool277/nginx",
"vendor": "nginx",
"versions": [
{
"status": "affected",
"version": "unspecified"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:nginx:nginx_http\\/2_module:*:*:*:*:*:*:*:*",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Ashrafov Jasur"
},
{
"lang": "en",
"type": "remediation developer",
"value": "Claude Opus 5.5"
}
],
"descriptions": [
{
"lang": "en",
"value": "A vulnerability exists in the HTTP/2 state machine of nginx (src/http/v2/ngx_http_v2.c). When a client sends an HTTP/2 header block that triggers a stream reset midway (for example, by including an invalid pseudo-header such as \":status\" in a request), the function ngx_http_v2_state_field_len() skips subsequent non-indexed literal fields without properly initializing state.field_start and state.field_end. The downstream function ngx_http_v2_state_process_header() then computes the header name and value lengths from these stale pointers. If the stale pointers happen to be equal (as on a connection where no literal field has yet been read), the skipped field is incorrectly rejected as a zero-length header name, and the entire HTTP/2 connection is terminated with GOAWAY(PROTOCOL_ERROR) rather than only the offending stream being reset. In other cases, skipped fields are charged the encoded length of a previously read field against the header list size limit, leading to incorrect resource accounting.\n\nAn unauthenticated remote attacker can exploit this by sending a single crafted HTTP/2 HEADERS frame containing an invalid pseudo-header followed by a literal header field. The result is the premature termination of the entire HTTP/2 connection, which can be repeated to cause a denial of service against the server\u0027s connection capacity. The secondary effect of incorrect header list size accounting could also be leveraged to bypass or distort the server\u0027s per-connection header size limits.\n\nThe fix ensures that when a field is skipped due to a stream reset, state.field_start is explicitly set to NULL and the actual encoded length is recorded in state.header. The process_header function then only derives name and value lengths from the field pointers when the field was actually read (field_start is non-NULL), preserving correct size accounting and preventing the false protocol-error condition that tore down the connection."
}
],
"impacts": [
{
"capecId": "CAPEC-107",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-107 State Manipulation"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "LOW",
"attackRequirements": "NONE",
"attackVector": "NETWORK",
"baseScore": 6.9,
"baseSeverity": "MEDIUM",
"privilegesRequired": "NONE",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "NONE",
"subIntegrityImpact": "NONE",
"userInteraction": "NONE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:L/SC:N/SI:N/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "LOW",
"vulnConfidentialityImpact": "NONE",
"vulnIntegrityImpact": "NONE",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-17T16:08:12Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-20",
"description": "CWE-20 Improper Input Validation",
"lang": "en",
"type": "CWE"
}
]
},
{
"descriptions": [
{
"cweId": "CWE-665",
"description": "CWE-665 Improper Initialization",
"lang": "en",
"type": "CWE"
}
]
}
],
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/kool277/nginx/commit/7b185de7a3f15e545888380cae25b81594f490e1"
}
],
"solutions": [
{
"lang": "en",
"value": "The fix corrects the HTTP/2 state machine so that when a stream is reset mid-header-block, subsequent literal fields are properly marked as skipped (state.field_start set to NULL) and their true encoded length is recorded in state.header. The header processing function now only computes name/value lengths from field pointers for fields that were actually read, preventing both the false zero-length rejection that caused connection-level GOAWAY and the incorrect header list size accounting for skipped fields."
}
],
"title": "nginx HTTP/2: Connection-level DoS via stale header field pointers after mid-block stream reset",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "none",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"The affected nginx version range is not specified in the patch; the commit references nginx/nginx#1977 but no version boundary is provided.",
"The CVSS VA:L rating assumes the impact is limited to a single HTTP/2 connection per attack instance; if the server has a very small connection pool, the practical DoS impact could be higher.",
"CAPEC-107 (State Manipulation) is the closest available CAPEC; no CAPEC specifically addresses HTTP/2 protocol state machine manipulation, so this mapping is approximate.",
"The secondary impact of incorrect header list size accounting (skipped fields charged with wrong lengths) is assessed as not independently exploitable for a distinct security impact beyond the primary DoS vector.",
"The patch is from a fork (kool277/nginx); it is assumed to correspond to an upstream nginx fix, but the exact upstream commit and version are not confirmed in the supplied metadata.",
"No evidence of authentication requirements, deployment topology, or multi-tenant context is present; the analysis assumes a standard single-tenant nginx HTTP/2 server."
],
"capecRationale": [
{
"capecId": "CAPEC-107",
"rationale": "The attacker sends a crafted HTTP/2 header block that triggers a stream reset (via an invalid pseudo-header) followed by a literal field, manipulating the server\u0027s internal HTTP/2 state machine into an inconsistent state where stale pointers cause a false protocol error and connection teardown. This is the closest CAPEC to the observed attack: the attacker exploits the application\u0027s state management logic by forcing a specific state transition sequence. Uncertainty: no CAPEC specifically covers HTTP/2 protocol state manipulation; CAPEC-107 is the best available match for exploiting state machine inconsistencies."
}
],
"commit": "7b185de7a3f15e545888380cae25b81594f490e1",
"confidence": "medium",
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Ashrafov Jasur"
},
{
"lang": "en",
"type": "remediation developer",
"value": "Claude Opus 5.5"
}
],
"cvssRationale": "AV:N: The vulnerability is exploitable over the network via a standard HTTP/2 connection. AC:L: The attack requires only sending a single crafted HEADERS frame with an invalid pseudo-header followed by a literal field; no race conditions or special timing are needed. AT:N: No manipulation of the attack target is required. PR:N: No authentication is needed to send HTTP/2 requests to a server. UI:N: No user interaction is required. VC:N, VI:N: No confidentiality or integrity impact is evident from the patch. VA:L: The impact is the termination of a single HTTP/2 connection (GOAWAY), which is a limited availability impact per instance; an attacker could repeat the attack across multiple connections but each instance only loses one connection. SC:N, SI:N, SA:N: No secondary component impact is indicated.",
"draft": true,
"fixSummary": "The fix corrects the HTTP/2 state machine so that when a stream is reset mid-header-block, subsequent literal fields are properly marked as skipped (state.field_start set to NULL) and their true encoded length is recorded in state.header. The header processing function now only computes name/value lengths from field pointers for fields that were actually read, preventing both the false zero-length rejection that caused connection-level GOAWAY and the incorrect header list size accounting for skipped fields.",
"generatedAt": "2026-11-17T16:08:12.480675Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 6,
"confidence": "medium",
"model": "qwen3.8:27b",
"score": 6
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "718197a690d2b6db645246443c0454c530412644b743df197d6fc92e34ba7a59",
"patchSummary": "In ngx_http_v2_state_field_len(), when the stream is NULL and the field is non-indexed (i.e., being skipped), the patch sets state.field_start to NULL and records the field\u0027s encoded length into state.header.name.len or state.header.value.len (with data set to NULL) before calling the skip handler. In ngx_http_v2_state_process_header(), the name and value length assignments from field_start/field_end are now guarded by a NULL check on state.field_start, so skipped fields retain the length recorded during the skip phase. A new regression test (tests/regression/h2_reset_stream_skipped_field.t) verifies that a stream reset followed by a literal field does not produce GOAWAY, that the connection remains usable, and that a skipped field with an empty name still triggers PROTOCOL_ERROR.",
"patchTruncated": false,
"patches": [
{
"commit": "7b185de7a3f15e545888380cae25b81594f490e1",
"date": "Sat, 14 Nov 2026 16:00:44 +0500",
"patchSha256": "718197a690d2b6db645246443c0454c530412644b743df197d6fc92e34ba7a59",
"retrievedAt": "2026-11-17T16:07:25.730817Z",
"source": "https://github.com/kool277/nginx/commit/7b185de7a3f15e545888380cae25b81594f490e1.patch",
"sourceUrl": "https://github.com/kool277/nginx/commit/7b185de7a3f15e545888380cae25b81594f490e1.patch",
"subject": "HTTP/2: fixed skipped header fields after a stream reset."
}
],
"source": "https://github.com/kool277/nginx/commit/7b185de7a3f15e545888380cae25b81594f490e1.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-17T16:08:12Z",
"version": "2.0.3"
},
"subject": "HTTP/2: fixed skipped header fields after a stream reset.",
"weaknessRationale": [
{
"cweId": "CWE-20",
"rationale": "The HTTP/2 state machine does not properly handle the transition from active field parsing to field skipping when a stream is reset mid-header-block. The stale pointer state leads to incorrect validation decisions (false zero-length rejection) and incorrect size accounting, both of which are consequences of failing to properly validate and track the state of skipped input."
},
{
"cweId": "CWE-665",
"rationale": "When entering the field-skip path, state.field_start and state.field_end are not re-initialized to reflect the skipped state. They retain values from a prior field, causing downstream logic to operate on stale data. The fix explicitly sets field_start to NULL to mark the field as skipped."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"description": "Corrective change authored (7b185de7a3f15e545888380cae25b81594f490e1): HTTP/2: fixed skipped header fields after a stream reset.",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/kool277/nginx/commit/7b185de7a3f15e545888380cae25b81594f490e1.patch"
],
"timestamp": "2026-11-14T11:00:44Z",
"type": "fix-developed"
},
{
"description": "Patch retrieved from the supplied URL; retrieval time is used as the default availability time.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/kool277/nginx/commit/7b185de7a3f15e545888380cae25b81594f490e1.patch"
],
"timestamp": "2026-11-17T16:07:25.730817Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20356"
}
]
}
},
"cveMetadata": {
"datePublished": "2026-11-17T16:10:37.898681Z",
"dateUpdated": "2026-11-17T16:10:37.898681Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20356"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2.0"
}
GCVE-1-2026-20338 (CVE-2026-129919)
Vulnerability from gna-1 – Published: 2026-11-16 13:48 – Updated: 2026-11-16 13:49| URL | Tags |
|---|---|
| https://github.com/kunai-project/kunai/commit/0ad… | patch |
| Vendor | Product | Version | |
|---|---|---|---|
| kunai-project | Kunai |
Affected:
0 , ≤ 0.6.2
(semver)
cpe:2.3:a:kunai-project:kunai:*:*:*:*:*:*:*:* |
qwen3.8:27b
2 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-16 13:44 - Model
qwen3.8:27b- Input
-
https://github.com/kunai-project/kunai/commit/0ad82911a6b72c81c55507fdabb9baad1ad1ad02.patch
1423cf59bf37… - Confidence
- medium
| Commit | Subject | Patch SHA-256 |
|---|---|---|
0ad82911a6b7
|
fix(namespace): verify opened mnt ns fd identity | 1423cf59bf37… |
Fix summary
The fix eliminates the TOCTOU race by opening the namespace file once to obtain a stable file descriptor, then reading the inode number from that fd via fstat/metadata and comparing it against the expected namespace's inode number. If the inodes do not match, the switcher creation is rejected with an UnexpectedNsInode error and nothing is cached, so the event is simply not enriched. This ensures the cached namespace identity is verified atomically from the opened fd rather than from a separate /proc lookup that could be affected by PID reuse or namespace changes.
Patch summary
In kunai/src/util/namespace.rs, Switcher::new now accepts an additional expected: &N parameter. Instead of calling N::from_pid to resolve the namespace (a separate /proc lookup), it calls N::open to get a stable fd, reads the inode via metadata().ino(), and compares it to expected.inum(). On mismatch it returns NsError::UnexpectedNsInode. A new NsError variant UnexpectedNsInode { cur, exp } is added. In kunai/src/cache.rs, the call site is updated to pass the expected namespace (&ns) to Switcher::new. Three new unit tests are added: one verifying that the cache rejects the host's own namespace when a different namespace is expected, one verifying rejection of a foreign namespace under the wrong key, and one verifying acceptance when the namespace matches.
CVSS rationale
AV:L: exploitation requires local access to create processes and manipulate namespaces. AC:H: requires a specific timing window between event generation and namespace resolution (PID reuse or setns call at the right moment). AT:P: the attacker must actively manipulate the target process (call setns) or arrange PID reuse. PR:L: local access is needed but not necessarily full root; setns requires appropriate capabilities but PID reuse is a normal system event. UI:N: no user interaction required. VC:N: no confidentiality impact. VI:L: integrity impact on the vulnerable component - security monitoring events are incorrectly resolved, producing wrong file paths. VA:N: no availability impact. SC:N/SI:L/SA:N: the security monitoring system (a separate component) produces incorrect integrity data (wrong path resolution) but no confidentiality or availability impact.
Weakness rationale
- CWE-367 The code performed a /proc/<pid>/ns/mnt lookup to determine the target namespace and then cached it under a different (expected) key. Between the event referencing the PID/namespace and the actual resolution, the PID could be reused or the process could change namespaces via setns/unshare, causing a different namespace to be cached under the wrong key. The fix verifies identity from a single opened fd, eliminating the race window.
- CWE-20 The original code did not validate that the resolved namespace matched the expected namespace before caching. The fix adds an explicit inode comparison and rejects the operation on mismatch, which is the missing input validation.
Attack pattern rationale
- CAPEC-110 The vulnerability is a classic TOCTOU race: the attacker exploits the time window between the event referencing a PID in a specific namespace and Kunai resolving that PID's actual namespace. By calling setns on the target process or arranging PID reuse within that window, the attacker causes a different namespace to be cached under the expected key. CAPEC-110 is the closest match. Uncertainty: the exact exploitation timing window is narrow and the patch does not describe a demonstrated exploit, so the mapping is based on the vulnerability class rather than a confirmed attack sequence.
Assumptions to verify
- The product is identified as Kunai from the repository path kunai-project/kunai; the vendor is not explicitly stated in the patch metadata and is recorded as unknown.
- No specific affected or fixed version numbers are provided in the patch; versions are recorded as unspecified.
- The CVSS PR:L assumes the attacker has local access with sufficient privileges to call setns or arrange PID reuse, but not necessarily full root. If CAP_SYS_ADMIN is strictly required, PR could be H.
- The CVSS AC:H reflects the narrow timing window required for the race condition; if PID reuse is considered a common enough event to lower the complexity, AC could be L.
- The CAPEC-110 mapping is based on the vulnerability class (TOCTOU race) rather than a demonstrated exploit; the patch does not include a proof-of-concept, so the exact attack sequence is inferred from the commit message description.
- The security impact is assessed as integrity-only (incorrect event resolution); the patch does not evidence any confidentiality or availability impact, but downstream effects on security operations are not fully characterized.
- The commit date is 2026-11-10; no CVE or advisory identifier is present in the patch metadata.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
5 | 11 | medium | 7 |
- 2026-11-10 08:53 UTC Fix developed Corrective change authored (0ad82911a6b72c81c55507fdabb9baad1ad1ad02): fix(namespace): verify opened mnt ns fd identity https://github.com/kunai-project/kunai/commit/0ad82911a6b72c81c55507fdabb9baad1ad1ad02.patch
- 2026-11-16 13:44 UTC Fix available Patch retrieved from the supplied URL; retrieval time is used as the default availability time. https://github.com/kunai-project/kunai/commit/0ad82911a6b72c81c55507fdabb9baad1ad1ad02.patch
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:kunai-project:kunai:*:*:*:*:*:*:*:*"
],
"modules": [
"kunai/src/cache.rs",
"kunai/src/util/namespace.rs"
],
"product": "Kunai",
"programFiles": [
"kunai/src/cache.rs",
"kunai/src/util/namespace.rs"
],
"repo": "https://github.com/kunai-project/kunai",
"vendor": "kunai-project",
"versions": [
{
"lessThanOrEqual": "0.6.2",
"status": "affected",
"version": "0",
"versionType": "semver"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:kunai-project:kunai:*:*:*:*:*:*:*:*",
"versionEndIncluding": "0.6.2",
"versionStartIncluding": "0",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "qjerome"
},
{
"lang": "en",
"type": "finder",
"value": "Wachizungu"
}
],
"descriptions": [
{
"lang": "en",
"value": "Kunai, a security monitoring tool, contains a time-of-check to time-of-use (TOCTOU) race condition in its mount namespace (mnt ns) caching logic. When resolving a security event that references a specific process PID and an expected mount namespace, the Switcher::new function read /proc/\u003cpid\u003e/ns/mnt to determine the target namespace and cached the result under the expected namespace key without verifying that the resolved namespace actually matched the expected one.\n\nAn attacker with local access could exploit this by either (1) arranging for the referenced PID to be reused by a process in a different namespace (e.g., the host namespace) before Kunai resolves it, or (2) calling unshare or setns on the target process after the event is generated but before the namespace is resolved. In either case, the wrong namespace (potentially the host\u0027s own namespace) would be cached under the container\u0027s namespace key.\n\nThe security impact is that subsequent container security events would be resolved against host filesystem paths rather than container paths. This undermines the integrity of the security monitoring data: container activity could be misattributed to the host, host activity could be misattributed to a container, and file-path-based detections or alerts would reference incorrect paths. An attacker could potentially use this to obscure malicious activity within a container by making it appear as benign host activity, or to generate misleading security alerts.\n\nThe vulnerability requires local access to the system and a specific timing window between event generation and namespace resolution. No remote attack vector or user interaction is required."
}
],
"impacts": [
{
"capecId": "CAPEC-110",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-110 Exploiting Race Conditions"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "HIGH",
"attackRequirements": "PRESENT",
"attackVector": "LOCAL",
"baseScore": 2,
"baseSeverity": "LOW",
"privilegesRequired": "LOW",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "NONE",
"subIntegrityImpact": "LOW",
"userInteraction": "NONE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:L/AC:H/AT:P/PR:L/UI:N/VC:N/VI:L/VA:N/SC:N/SI:L/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "NONE",
"vulnConfidentialityImpact": "NONE",
"vulnIntegrityImpact": "LOW",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "no"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-16T13:44:48Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-367",
"description": "CWE-367 TOCTOU Race Condition",
"lang": "en",
"type": "CWE"
}
]
},
{
"descriptions": [
{
"cweId": "CWE-20",
"description": "CWE-20 Improper Input Validation",
"lang": "en",
"type": "CWE"
}
]
}
],
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/kunai-project/kunai/commit/0ad82911a6b72c81c55507fdabb9baad1ad1ad02"
}
],
"solutions": [
{
"lang": "en",
"value": "The fix eliminates the TOCTOU race by opening the namespace file once to obtain a stable file descriptor, then reading the inode number from that fd via fstat/metadata and comparing it against the expected namespace\u0027s inode number. If the inodes do not match, the switcher creation is rejected with an UnexpectedNsInode error and nothing is cached, so the event is simply not enriched. This ensures the cached namespace identity is verified atomically from the opened fd rather than from a separate /proc lookup that could be affected by PID reuse or namespace changes."
}
],
"title": "Kunai Mount Namespace Cache TOCTOU Allows Incorrect Event Resolution Against Host Files",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "full",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"The product is identified as Kunai from the repository path kunai-project/kunai; the vendor is not explicitly stated in the patch metadata and is recorded as unknown.",
"No specific affected or fixed version numbers are provided in the patch; versions are recorded as unspecified.",
"The CVSS PR:L assumes the attacker has local access with sufficient privileges to call setns or arrange PID reuse, but not necessarily full root. If CAP_SYS_ADMIN is strictly required, PR could be H.",
"The CVSS AC:H reflects the narrow timing window required for the race condition; if PID reuse is considered a common enough event to lower the complexity, AC could be L.",
"The CAPEC-110 mapping is based on the vulnerability class (TOCTOU race) rather than a demonstrated exploit; the patch does not include a proof-of-concept, so the exact attack sequence is inferred from the commit message description.",
"The security impact is assessed as integrity-only (incorrect event resolution); the patch does not evidence any confidentiality or availability impact, but downstream effects on security operations are not fully characterized.",
"The commit date is 2026-11-10; no CVE or advisory identifier is present in the patch metadata."
],
"capecRationale": [
{
"capecId": "CAPEC-110",
"rationale": "The vulnerability is a classic TOCTOU race: the attacker exploits the time window between the event referencing a PID in a specific namespace and Kunai resolving that PID\u0027s actual namespace. By calling setns on the target process or arranging PID reuse within that window, the attacker causes a different namespace to be cached under the expected key. CAPEC-110 is the closest match. Uncertainty: the exact exploitation timing window is narrow and the patch does not describe a demonstrated exploit, so the mapping is based on the vulnerability class rather than a confirmed attack sequence."
}
],
"commit": "0ad82911a6b72c81c55507fdabb9baad1ad1ad02",
"confidence": "medium",
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "qjerome"
}
],
"cvssRationale": "AV:L: exploitation requires local access to create processes and manipulate namespaces. AC:H: requires a specific timing window between event generation and namespace resolution (PID reuse or setns call at the right moment). AT:P: the attacker must actively manipulate the target process (call setns) or arrange PID reuse. PR:L: local access is needed but not necessarily full root; setns requires appropriate capabilities but PID reuse is a normal system event. UI:N: no user interaction required. VC:N: no confidentiality impact. VI:L: integrity impact on the vulnerable component - security monitoring events are incorrectly resolved, producing wrong file paths. VA:N: no availability impact. SC:N/SI:L/SA:N: the security monitoring system (a separate component) produces incorrect integrity data (wrong path resolution) but no confidentiality or availability impact.",
"draft": false,
"fixSummary": "The fix eliminates the TOCTOU race by opening the namespace file once to obtain a stable file descriptor, then reading the inode number from that fd via fstat/metadata and comparing it against the expected namespace\u0027s inode number. If the inodes do not match, the switcher creation is rejected with an UnexpectedNsInode error and nothing is cached, so the event is simply not enriched. This ensures the cached namespace identity is verified atomically from the opened fd rather than from a separate /proc lookup that could be affected by PID reuse or namespace changes.",
"generatedAt": "2026-11-16T13:44:48.393000Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 7,
"confidence": "medium",
"model": "qwen3.8:27b",
"score": 5
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "1423cf59bf3736f481f429f0e482e25af19ec53348a90d9ce265b254ba198576",
"patchSummary": "In kunai/src/util/namespace.rs, Switcher::new now accepts an additional expected: \u0026N parameter. Instead of calling N::from_pid to resolve the namespace (a separate /proc lookup), it calls N::open to get a stable fd, reads the inode via metadata().ino(), and compares it to expected.inum(). On mismatch it returns NsError::UnexpectedNsInode. A new NsError variant UnexpectedNsInode { cur, exp } is added. In kunai/src/cache.rs, the call site is updated to pass the expected namespace (\u0026ns) to Switcher::new. Three new unit tests are added: one verifying that the cache rejects the host\u0027s own namespace when a different namespace is expected, one verifying rejection of a foreign namespace under the wrong key, and one verifying acceptance when the namespace matches.",
"patchTruncated": false,
"patches": [
{
"commit": "0ad82911a6b72c81c55507fdabb9baad1ad1ad02",
"date": "Tue, 10 Nov 2026 09:53:52 +0100",
"patchSha256": "1423cf59bf3736f481f429f0e482e25af19ec53348a90d9ce265b254ba198576",
"retrievedAt": "2026-11-16T13:44:10.582832Z",
"source": "https://github.com/kunai-project/kunai/commit/0ad82911a6b72c81c55507fdabb9baad1ad1ad02.patch",
"sourceUrl": "https://github.com/kunai-project/kunai/commit/0ad82911a6b72c81c55507fdabb9baad1ad1ad02.patch",
"subject": "fix(namespace): verify opened mnt ns fd identity"
}
],
"source": "https://github.com/kunai-project/kunai/commit/0ad82911a6b72c81c55507fdabb9baad1ad1ad02.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "no"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-16T13:44:48Z",
"version": "2.0.3"
},
"subject": "fix(namespace): verify opened mnt ns fd identity",
"weaknessRationale": [
{
"cweId": "CWE-367",
"rationale": "The code performed a /proc/\u003cpid\u003e/ns/mnt lookup to determine the target namespace and then cached it under a different (expected) key. Between the event referencing the PID/namespace and the actual resolution, the PID could be reused or the process could change namespaces via setns/unshare, causing a different namespace to be cached under the wrong key. The fix verifies identity from a single opened fd, eliminating the race window."
},
{
"cweId": "CWE-20",
"rationale": "The original code did not validate that the resolved namespace matched the expected namespace before caching. The fix adds an explicit inode comparison and rejects the operation on mismatch, which is the missing input validation."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"description": "Corrective change authored (0ad82911a6b72c81c55507fdabb9baad1ad1ad02): fix(namespace): verify opened mnt ns fd identity",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/kunai-project/kunai/commit/0ad82911a6b72c81c55507fdabb9baad1ad1ad02.patch"
],
"timestamp": "2026-11-10T08:53:52Z",
"type": "fix-developed"
},
{
"description": "Patch retrieved from the supplied URL; retrieval time is used as the default availability time.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/kunai-project/kunai/commit/0ad82911a6b72c81c55507fdabb9baad1ad1ad02.patch"
],
"timestamp": "2026-11-16T13:44:10.582832Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20338"
}
]
}
},
"cveMetadata": {
"cveId": "CVE-2026-129919",
"datePublished": "2026-11-16T13:48:19.935482Z",
"dateReserved": "2026-11-16T13:49:40.560Z",
"dateUpdated": "2026-11-16T13:49:41.771044Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20338"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2.0"
}
GCVE-1-2026-20340 (CVE-2026-129706)
Vulnerability from gna-1 – Published: 2026-11-16 08:44 – Updated: 2026-11-16 08:46- CWE-613 - Session Persistence After Password Change
| URL | Tags |
|---|---|
| https://github.com/vulnerability-lookup/vulnerabi… | patch |
| Vendor | Product | Version | |
|---|---|---|---|
| vulnerability-lookup | vulnerability-lookup |
Affected:
unspecified
cpe:2.3:a:vulnerability-lookup:vulnerability-lookup:*:*:*:*:*:*:*:* |
qwen3.8:27b
4 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-16 08:39 - Model
qwen3.8:27b- Input
-
https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d.patch
e35954004742… - Confidence
- medium
| Commit | Subject | Patch SHA-256 |
|---|---|---|
d762810fb3b9
|
fix: [security] A password change ends the user's other | e35954004742… |
Fix summary
The session identifier is now bound to the user's current password hash. User.get_id() returns a composite string of the form '<user_id>:<digest>', where the digest is an HMAC-SHA256 of the stored password hash keyed with the application SECRET_KEY. The load_user callback validates that the digest in the session matches the digest computed from the user's current password hash using a constant-time comparison (hmac.compare_digest). If the password has changed, the digest no longer matches and the session is rejected, effectively logging out all sessions that were opened under the previous credential. When a user changes their own password through the profile page, they are immediately re-authenticated via login_user() so their current session is renewed with the new digest while all other sessions are invalidated.
Patch summary
In website/models/user.py, a new session_digest() method computes an HMAC-SHA256 of the password hash keyed with SECRET_KEY and returns the first 32 hex characters. User.get_id() is changed from returning the integer self.id to returning the string f'{self.id}:{self.session_digest()}'. In website/web/views/session_mgmt.py, the load_user callback is rewritten to parse the composite session ID, extract the user ID and digest, look up the user, and verify the digest with hmac.compare_digest against the user's current session_digest(); sessions with a bare ID or a mismatched digest are rejected. In website/web/views/user.py, the ineffective session.new = True assignment is removed and replaced with a call to login_user(user) after a password change so the user's own session is renewed. All test files are updated to use user.get_id() instead of user.id when setting the session, and a new test file tests/test_session_password_binding.py verifies that a password change invalidates other sessions, that bare IDs are拒绝
CVSS rationale
AV:N: the application is a network-accessible web service. AC:L: once a valid session is obtained, it persists with no additional complexity; no race conditions or special timing are needed. AT:N: no attack target manipulation is required. PR:L: the attacker must already possess a valid authenticated session for the target user (obtained via XSS, session hijacking, or a previously exposed key). UI:N: no victim interaction is needed after the session is compromised. VC:L: the attacker can read the user's profile data, subscriptions, and role-gated content. VI:L: the attacker can modify the user's profile, subscriptions, and other user-controlled data. VA:L: the attacker can continue to use the service on behalf of the user, and the user cannot revoke access by changing the password. SC/SI/SA:N: no impact on other systems or components is evidenced in the patch.
Weakness rationale
- CWE-613 The commit message explicitly cites CWE-613. The session ID was the bare user ID with no component tied to the credential, so sessions outlived every password change, account recovery, or admin reset. This is a textbook instance of session persistence after password change.
Attack pattern rationale
- CAPEC-122 The session identifier was effectively fixed to the static user ID and never rotated or invalidated on credential change. While classic session fixation involves an attacker choosing the session ID, the underlying weakness is the same: the session ID is not bound to the current credential state and cannot be revoked by the user. CAPEC-122 is the closest available pattern; the uncertainty is that the attacker does not choose the ID but rather exploits its permanence. No CAPEC specifically covers 'session outlives credential change' as a distinct pattern, making CAPEC-122 the best available match.
Assumptions to verify
- The application is assumed to be a standard network-accessible Flask web application; no specific deployment topology, reverse proxy, or authentication gateway is evidenced in the patch.
- PR:L assumes the attacker must already hold a valid session for the target user. If sessions could be obtained without authentication (e.g., via a feed key or API key that does not require a prior login), PR could be N; the commit message references feed keys and API keys as a vector, but the exact authentication requirements of those keys are not detailed in the patch.
- The CAPEC-122 mapping is the closest available pattern; no CAPEC specifically describes 'session outlives credential change' as a standalone attack pattern, so the mapping carries inherent uncertainty.
- Affected version range is unspecified because the patch does not include version tags, release notes, or a changelog entry identifying the first or last affected release.
- The commit date is 2026-11-16; the fixed version is the commit d762810fb3b9d1ee34938a71cd8cbee544c1527d and all prior commits that lacked the digest-bound session ID.
- CVSS impact metrics (VC, VI, VA) are rated Low rather than High because the attacker's capabilities are bounded by the permissions the session already carried; the vulnerability does not grant new privileges but prevents revocation of existing ones.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
6 | 11 | medium | 6 |
- 2026-11-15 16:36 UTC Report received maintainer Vulnerability report received by e-mail from the reporter (Michael Reimsbach), against revision 4f181ad9 on main. Besides two token findings, it reports that an existing session is not invalidated when the password or a key changes (CWE-613).
- 2026-11-16 07:40 UTC Validated maintainer Report reproduced on revision 4f181ad9: a session stays valid after the user's password, API key and feed key are all changed, because the session identifier is the bare user ID.
- 2026-11-16 08:04 UTC Fix developed maintainer Corrective change authored (d762810fb3b9d1ee34938a71cd8cbee544c1527d): fix: [security] A password change ends the user's other sessions. The session identifier is bound to the password hash; sessions holding a bare ID are refused, which also ends the sessions a feed key or an API key could open before 3605d5b6. https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d.patch
- 2026-11-16 08:06 UTC Fix available maintainer Fix pushed to the public main branch of the repository, together with 3605d5b6. https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d.patch
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:vulnerability-lookup:vulnerability-lookup:*:*:*:*:*:*:*:*"
],
"modules": [
"website/models/user.py",
"website/web/views/session_mgmt.py",
"website/web/views/user.py"
],
"product": "vulnerability-lookup",
"programFiles": [
"website/models/user.py",
"website/web/views/session_mgmt.py",
"website/web/views/user.py"
],
"repo": "https://github.com/vulnerability-lookup/vulnerability-lookup",
"vendor": "vulnerability-lookup",
"versions": [
{
"status": "affected",
"version": "unspecified"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:vulnerability-lookup:vulnerability-lookup:*:*:*:*:*:*:*:*",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "reporter",
"value": "Michael Reimsbach"
},
{
"lang": "en",
"type": "remediation developer",
"value": "C\u00e9dric Bonhomme"
},
{
"lang": "en",
"type": "remediation developer",
"value": "Claude Opus 5.5"
}
],
"descriptions": [
{
"lang": "en",
"value": "In the vulnerability-lookup web application, the Flask-Login session identifier was set to the bare numeric user ID (User.get_id() returned self.id). Because the session ID contained no component derived from the user\u0027s credential, it remained valid indefinitely regardless of any subsequent password change, account recovery, or administrator-initiated password reset. An attacker who had obtained a valid session for a user account (for example through cross-site scripting, session hijacking, or a previously exposed feed or API key) could continue to access that user\u0027s account after the user changed their password, because the session cookie still carried the same static user ID and the user_loader accepted it without any credential-bound validation.\n\nThe affected component is the session management layer of the vulnerability-lookup Flask application, specifically the User.get_id() method in website/models/user.py and the load_user callback in website/web/views/session_mgmt.py. The vulnerability required that an attacker already possessed a valid session for the target user; no additional authentication or special deployment configuration was needed beyond a standard deployment of the application.\n\nThe security impact is that a user cannot effectively revoke an attacker\u0027s access by changing their password, performing an account recovery, or having an administrator reset the credential. The attacker retains full authenticated access to the user\u0027s profile, subscriptions, and any role-gated functionality for the lifetime of the session cookie. Additionally, sessions that were opened via a feed key or API key before a prior fix (commit 3605d5b6) also persisted indefinitely."
}
],
"impacts": [
{
"capecId": "CAPEC-122",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-122 Session Fixation"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "LOW",
"attackRequirements": "NONE",
"attackVector": "NETWORK",
"baseScore": 5.3,
"baseSeverity": "MEDIUM",
"privilegesRequired": "LOW",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "NONE",
"subIntegrityImpact": "NONE",
"userInteraction": "NONE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:L/VI:L/VA:L/SC:N/SI:N/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "LOW",
"vulnConfidentialityImpact": "LOW",
"vulnIntegrityImpact": "LOW",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-16T08:39:36Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-613",
"description": "CWE-613 Session Persistence After Password Change",
"lang": "en",
"type": "CWE"
}
]
}
],
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d"
}
],
"solutions": [
{
"lang": "en",
"value": "The session identifier is now bound to the user\u0027s current password hash. User.get_id() returns a composite string of the form \u0027\u003cuser_id\u003e:\u003cdigest\u003e\u0027, where the digest is an HMAC-SHA256 of the stored password hash keyed with the application SECRET_KEY. The load_user callback validates that the digest in the session matches the digest computed from the user\u0027s current password hash using a constant-time comparison (hmac.compare_digest). If the password has changed, the digest no longer matches and the session is rejected, effectively logging out all sessions that were opened under the previous credential. When a user changes their own password through the profile page, they are immediately re-authenticated via login_user() so their current session is renewed with the new digest while all other sessions are invalidated."
}
],
"title": "vulnerability-lookup: Session Persists After Password Change (CWE-613)",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "partial",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"The application is assumed to be a standard network-accessible Flask web application; no specific deployment topology, reverse proxy, or authentication gateway is evidenced in the patch.",
"PR:L assumes the attacker must already hold a valid session for the target user. If sessions could be obtained without authentication (e.g., via a feed key or API key that does not require a prior login), PR could be N; the commit message references feed keys and API keys as a vector, but the exact authentication requirements of those keys are not detailed in the patch.",
"The CAPEC-122 mapping is the closest available pattern; no CAPEC specifically describes \u0027session outlives credential change\u0027 as a standalone attack pattern, so the mapping carries inherent uncertainty.",
"Affected version range is unspecified because the patch does not include version tags, release notes, or a changelog entry identifying the first or last affected release.",
"The commit date is 2026-11-16; the fixed version is the commit d762810fb3b9d1ee34938a71cd8cbee544c1527d and all prior commits that lacked the digest-bound session ID.",
"CVSS impact metrics (VC, VI, VA) are rated Low rather than High because the attacker\u0027s capabilities are bounded by the permissions the session already carried; the vulnerability does not grant new privileges but prevents revocation of existing ones."
],
"capecRationale": [
{
"capecId": "CAPEC-122",
"rationale": "The session identifier was effectively fixed to the static user ID and never rotated or invalidated on credential change. While classic session fixation involves an attacker choosing the session ID, the underlying weakness is the same: the session ID is not bound to the current credential state and cannot be revoked by the user. CAPEC-122 is the closest available pattern; the uncertainty is that the attacker does not choose the ID but rather exploits its permanence. No CAPEC specifically covers \u0027session outlives credential change\u0027 as a distinct pattern, making CAPEC-122 the best available match."
}
],
"commit": "d762810fb3b9d1ee34938a71cd8cbee544c1527d",
"confidence": "medium",
"credits": [
{
"lang": "en",
"type": "reporter",
"value": "Michael Reimsbach"
},
{
"lang": "en",
"type": "remediation developer",
"value": "C\u00e9dric Bonhomme"
},
{
"lang": "en",
"type": "remediation developer",
"value": "Claude Opus 5.5"
}
],
"cvssRationale": "AV:N: the application is a network-accessible web service. AC:L: once a valid session is obtained, it persists with no additional complexity; no race conditions or special timing are needed. AT:N: no attack target manipulation is required. PR:L: the attacker must already possess a valid authenticated session for the target user (obtained via XSS, session hijacking, or a previously exposed key). UI:N: no victim interaction is needed after the session is compromised. VC:L: the attacker can read the user\u0027s profile data, subscriptions, and role-gated content. VI:L: the attacker can modify the user\u0027s profile, subscriptions, and other user-controlled data. VA:L: the attacker can continue to use the service on behalf of the user, and the user cannot revoke access by changing the password. SC/SI/SA:N: no impact on other systems or components is evidenced in the patch.",
"draft": false,
"fixSummary": "The session identifier is now bound to the user\u0027s current password hash. User.get_id() returns a composite string of the form \u0027\u003cuser_id\u003e:\u003cdigest\u003e\u0027, where the digest is an HMAC-SHA256 of the stored password hash keyed with the application SECRET_KEY. The load_user callback validates that the digest in the session matches the digest computed from the user\u0027s current password hash using a constant-time comparison (hmac.compare_digest). If the password has changed, the digest no longer matches and the session is rejected, effectively logging out all sessions that were opened under the previous credential. When a user changes their own password through the profile page, they are immediately re-authenticated via login_user() so their current session is renewed with the new digest while all other sessions are invalidated.",
"generatedAt": "2026-11-16T08:39:36.754806Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 6,
"confidence": "medium",
"model": "qwen3.8:27b",
"score": 6
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "e359540047428f45208a6edd589f39b561414e337fbaf9fcba7f34d63e1dae0f",
"patchSummary": "In website/models/user.py, a new session_digest() method computes an HMAC-SHA256 of the password hash keyed with SECRET_KEY and returns the first 32 hex characters. User.get_id() is changed from returning the integer self.id to returning the string f\u0027{self.id}:{self.session_digest()}\u0027. In website/web/views/session_mgmt.py, the load_user callback is rewritten to parse the composite session ID, extract the user ID and digest, look up the user, and verify the digest with hmac.compare_digest against the user\u0027s current session_digest(); sessions with a bare ID or a mismatched digest are rejected. In website/web/views/user.py, the ineffective session.new = True assignment is removed and replaced with a call to login_user(user) after a password change so the user\u0027s own session is renewed. All test files are updated to use user.get_id() instead of user.id when setting the session, and a new test file tests/test_session_password_binding.py verifies that a password change invalidates other sessions, that bare IDs are\u62d2\u7edd",
"patchTruncated": false,
"patches": [
{
"commit": "d762810fb3b9d1ee34938a71cd8cbee544c1527d",
"date": "Mon, 16 Nov 2026 09:04:41 +0100",
"patchSha256": "e359540047428f45208a6edd589f39b561414e337fbaf9fcba7f34d63e1dae0f",
"retrievedAt": "2026-11-16T08:38:48.822217Z",
"source": "https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d.patch",
"sourceUrl": "https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d.patch",
"subject": "fix: [security] A password change ends the user\u0027s other"
}
],
"source": "https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-16T08:39:36Z",
"version": "2.0.3"
},
"subject": "fix: [security] A password change ends the user\u0027s other",
"weaknessRationale": [
{
"cweId": "CWE-613",
"rationale": "The commit message explicitly cites CWE-613. The session ID was the bare user ID with no component tied to the credential, so sessions outlived every password change, account recovery, or admin reset. This is a textbook instance of session persistence after password change."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"actor": "maintainer",
"description": "Vulnerability report received by e-mail from the reporter (Michael Reimsbach), against revision 4f181ad9 on main. Besides two token findings, it reports that an existing session is not invalidated when the password or a key changes (CWE-613).",
"id": "evt-received-1",
"timestamp": "2026-11-15T16:36:10Z",
"type": "received"
},
{
"actor": "maintainer",
"description": "Report reproduced on revision 4f181ad9: a session stays valid after the user\u0027s password, API key and feed key are all changed, because the session identifier is the bare user ID.",
"id": "evt-validated-1",
"timestamp": "2026-11-16T07:40:02Z",
"type": "validated"
},
{
"actor": "maintainer",
"description": "Corrective change authored (d762810fb3b9d1ee34938a71cd8cbee544c1527d): fix: [security] A password change ends the user\u0027s other sessions. The session identifier is bound to the password hash; sessions holding a bare ID are refused, which also ends the sessions a feed key or an API key could open before 3605d5b6.",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d",
"https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d.patch"
],
"timestamp": "2026-11-16T08:04:41Z",
"type": "fix-developed"
},
{
"actor": "maintainer",
"description": "Fix pushed to the public main branch of the repository, together with 3605d5b6.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d",
"https://github.com/vulnerability-lookup/vulnerability-lookup/commit/d762810fb3b9d1ee34938a71cd8cbee544c1527d.patch"
],
"timestamp": "2026-11-16T08:06:43Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20340"
}
]
}
},
"cveMetadata": {
"cveId": "CVE-2026-129706",
"datePublished": "2026-11-16T08:44:56.097997Z",
"dateReserved": "2026-11-16T08:46:40.876Z",
"dateUpdated": "2026-11-16T08:46:41.306533Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20340"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2.0"
}
GCVE-1-2026-20245 (CVE-2026-129703)
Vulnerability from gna-1 – Published: 2026-11-16 08:31 – Updated: 2026-11-16 08:36| URL | Tags |
|---|---|
| https://github.com/vulnerability-lookup/vulnerabi… | patch |
| Vendor | Product | Version | |
|---|---|---|---|
| vulnerability-lookup | vulnerability-lookup |
Affected:
unspecified
cpe:2.3:a:vulnerability-lookup:vulnerability-lookup:*:*:*:*:*:*:*:* |
qwen3.8:27b
4 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-16 08:23 - Model
qwen3.8:27b- Input
-
https://github.com/vulnerability-lookup/vulnerability-lookup/commit/3605d5b65d340343cc4b612d40d454afed84302b.patch
36e0430891b6… - Confidence
- high
| Commit | Subject | Patch SHA-256 |
|---|---|---|
3605d5b65d34
|
fix: [security] A feed key or an API key no longer opens a | 36e0430891b6… |
Fix summary
The authentication decorators for feed-key and API-key requests no longer call Flask-Login's login_user(), which was writing the authenticated user into the session cookie and creating a full web session. Instead, the user object is now stored on Flask's request-scoped g._login_user attribute, which Flask-Login's current_user property reads first. This authenticates only the current request without setting any session cookie, preventing a limited-scope token from being elevated to a persistent, all-pages session that bypasses password and two-factor authentication.
Patch summary
In website/decorators.py (auth_feedkey_func) and website/web/api/v1/common.py (auth_func), the import of login_user from flask_login is removed and replaced with an import of g from flask. The call login_user(user) (or login_user(user, remember=False)) is replaced with g._login_user = user. A new test file tests/test_token_auth_no_session.py is added with four test cases verifying that: (1) a feed-key request returns 200 but sets no session cookie, (2) an API-key request returns 200 but sets no session cookie, (3) a subsequent unauthenticated request to /user/profile is redirected (302), and (4) a 403 response from an API-key request also sets no session cookie.
CVSS rationale
AV:N: The feed key is in a URL and the API key is in a header; both are accessible over the network. AC:L: A single HTTP request with the token is sufficient; no race conditions or special timing required. AT:N: No manipulation of the target system is needed. PR:N: No prior authentication is required; the token itself is the credential, and it is not protected by any additional gate. UI:N: No user interaction is needed. VC/VI/VA:N: The vulnerable component (the authentication layer) is not itself compromised in terms of its own data, integrity, or availability. SC:H: The attacker gains full read access to the user's account data, including the API key and profile information. SI:H: The attacker can modify the user's account settings, and for administrator accounts, can perform arbitrary administrative actions via /admin. SA:N: No availability impact on the user's account or the application is demonstrated by the evidence.
Weakness rationale
- CWE-287 The authentication mechanism improperly elevated a limited-scope credential (feed key or API key intended for single-request authentication) into a full web session via login_user(), bypassing password and two-factor authentication. The authentication check succeeded but the session establishment was incorrect for the credential type.
- CWE-613 The commit message explicitly references CWE-613 as a related concern: sessions created before the fix remain valid until they expire and are not bound to the original credential. While the primary fix addresses the session creation, the lack of credential binding in existing sessions is a secondary weakness acknowledged by the developers.
Attack pattern rationale
- CAPEC-63 The feed key and API key are trusted credentials for specific, limited-scope access (feed delivery, API calls). The vulnerable implementation incorrectly leveraged that trust to establish a full web session, extending the credential's authority beyond its intended scope. This is the closest CAPEC to the pattern of a limited credential being used to gain broader access than intended. Uncertainty: no CAPEC precisely describes 'token authentication creating an unintended persistent session'; CAPEC-63 is the best available match for trust-scope escalation.
Assumptions to verify
- The application is assumed to be a self-hosted or publicly accessible web service; the patch does not specify deployment topology.
- No specific version numbers are available from the patch; the affected version range is unspecified.
- The CVSS sub-component impact (SC:H, SI:H) assumes the worst case where the authenticated account has administrative privileges, granting access to /admin. For non-admin accounts, the impact is limited to the user's own account data.
- The CAPEC-63 mapping is the closest available match; no CAPEC precisely describes a limited-scope token being elevated to a full session via an improper login_user() call.
- The commit message notes that pre-existing sessions created before the fix remain valid until expiration and are not bound to the credential (CWE-613); this is acknowledged as a separate issue not addressed by this patch.
- The 'Co-Authored-By: Claude Opus 5.5' credit is included as a tool credit based on the explicit commit message; it is unclear whether this represents meaningful remediation contribution or automated assistance.
- The patch does not indicate whether the application requires TLS or any other transport-level protection; CVSS assumes standard network access.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
6 | 11 | high | 7 |
- 2026-11-15 16:36 UTC Report received maintainer Vulnerability report received by e-mail from the reporter.
- 2026-11-16 07:40 UTC Validated maintainer Report reproduced: a feed key or an API key opens a full web session, which also bypasses two-factor authentication.
- 2026-11-16 07:56 UTC Fix developed maintainer Corrective change authored (3605d5b6): a feed key or an API key no longer opens a web session. https://github.com/vulnerability-lookup/vulnerability-lookup/commit/3605d5b65d340343cc4b612d40d454afed84302b
- 2026-11-16 08:06 UTC Fix available maintainer Fix pushed to the public main branch. https://github.com/vulnerability-lookup/vulnerability-lookup/commit/3605d5b65d340343cc4b612d40d454afed84302b
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:vulnerability-lookup:vulnerability-lookup:*:*:*:*:*:*:*:*"
],
"modules": [
"website/decorators.py (auth_feedkey_func)",
"website/web/api/v1/common.py (auth_func)"
],
"product": "vulnerability-lookup",
"programFiles": [
"website/decorators.py",
"website/web/api/v1/common.py"
],
"repo": "https://github.com/vulnerability-lookup/vulnerability-lookup",
"vendor": "vulnerability-lookup",
"versions": [
{
"status": "affected",
"version": "unspecified"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:vulnerability-lookup:vulnerability-lookup:*:*:*:*:*:*:*:*",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "reporter",
"value": "Michael Reimsbach"
},
{
"lang": "en",
"type": "remediation developer",
"value": "C\u00e9dric Bonhomme"
},
{
"lang": "en",
"type": "remediation developer",
"value": "Claude Opus 5.5"
},
{
"lang": "en",
"type": "tool",
"value": "Claude Opus 5.5"
}
],
"descriptions": [
{
"lang": "en",
"value": "In the vulnerability-lookup web application, the authentication decorators for feed-key requests (auth_feedkey_func in website/decorators.py) and API-key requests (auth_func in website/web/api/v1/common.py) called Flask-Login\u0027s login_user() function. This caused the HTTP response to a request authenticated solely by a feed key (supplied as the X-FEED-KEY URL parameter) or an API key (supplied as the X-API-KEY header) to set a session cookie that grants full authenticated access to every page in the application, without requiring a password or two-factor authentication.\n\nThe practical impact is that any party possessing a leaked feed URL (a common occurrence, as feed URLs are shared with RSS readers) or an API key could obtain a valid web session for the associated account. This session provides access to the user\u0027s account page, which exposes the API key and other profile data, and for accounts with the administrator role, access to the /admin interface, enabling full administrative control of the application.\n\nThe vulnerability affects all users whose feed key or API key is exposed. No prior authentication, user interaction, or special network conditions are required; a single HTTP request with the token is sufficient to establish the session. The session persists until it expires, and because it is not bound to the original credential (a related CWE-613 issue noted by the developers), revoking the feed key or API key does not invalidate the already-created session.\n\nThe fix replaces the login_user() call with assignment to g._login_user, which Flask-Login\u0027s current_user property reads for the current request only, without writing any session cookie. The affected views remain unchanged and continue to see the authenticated user for the single request."
}
],
"impacts": [
{
"capecId": "CAPEC-63",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-63 Leverage Trusted Relationships"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "LOW",
"attackRequirements": "NONE",
"attackVector": "NETWORK",
"baseScore": 7.8,
"baseSeverity": "HIGH",
"privilegesRequired": "NONE",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "HIGH",
"subIntegrityImpact": "HIGH",
"userInteraction": "NONE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:N/VI:N/VA:N/SC:H/SI:H/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "NONE",
"vulnConfidentialityImpact": "NONE",
"vulnIntegrityImpact": "NONE",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "total"
}
],
"role": "Supplier",
"timestamp": "2026-11-16T08:23:33Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-287",
"description": "CWE-287 Improper Authentication",
"lang": "en",
"type": "CWE"
}
]
},
{
"descriptions": [
{
"cweId": "CWE-613",
"description": "CWE-613 Insufficient Session Expiration",
"lang": "en",
"type": "CWE"
}
]
}
],
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/vulnerability-lookup/vulnerability-lookup/commit/3605d5b65d340343cc4b612d40d454afed84302b"
}
],
"solutions": [
{
"lang": "en",
"value": "The authentication decorators for feed-key and API-key requests no longer call Flask-Login\u0027s login_user(), which was writing the authenticated user into the session cookie and creating a full web session. Instead, the user object is now stored on Flask\u0027s request-scoped g._login_user attribute, which Flask-Login\u0027s current_user property reads first. This authenticates only the current request without setting any session cookie, preventing a limited-scope token from being elevated to a persistent, all-pages session that bypasses password and two-factor authentication."
}
],
"title": "vulnerability-lookup: Feed key or API key authentication creates full web session bypassing password and 2FA",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "partial",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"The application is assumed to be a self-hosted or publicly accessible web service; the patch does not specify deployment topology.",
"No specific version numbers are available from the patch; the affected version range is unspecified.",
"The CVSS sub-component impact (SC:H, SI:H) assumes the worst case where the authenticated account has administrative privileges, granting access to /admin. For non-admin accounts, the impact is limited to the user\u0027s own account data.",
"The CAPEC-63 mapping is the closest available match; no CAPEC precisely describes a limited-scope token being elevated to a full session via an improper login_user() call.",
"The commit message notes that pre-existing sessions created before the fix remain valid until expiration and are not bound to the credential (CWE-613); this is acknowledged as a separate issue not addressed by this patch.",
"The \u0027Co-Authored-By: Claude Opus 5.5\u0027 credit is included as a tool credit based on the explicit commit message; it is unclear whether this represents meaningful remediation contribution or automated assistance.",
"The patch does not indicate whether the application requires TLS or any other transport-level protection; CVSS assumes standard network access."
],
"capecRationale": [
{
"capecId": "CAPEC-63",
"rationale": "The feed key and API key are trusted credentials for specific, limited-scope access (feed delivery, API calls). The vulnerable implementation incorrectly leveraged that trust to establish a full web session, extending the credential\u0027s authority beyond its intended scope. This is the closest CAPEC to the pattern of a limited credential being used to gain broader access than intended. Uncertainty: no CAPEC precisely describes \u0027token authentication creating an unintended persistent session\u0027; CAPEC-63 is the best available match for trust-scope escalation."
}
],
"commit": "3605d5b65d340343cc4b612d40d454afed84302b",
"confidence": "high",
"credits": [
{
"lang": "en",
"type": "reporter",
"value": "Michael Reimsbach"
},
{
"lang": "en",
"type": "remediation developer",
"value": "C\u00e9dric Bonhomme"
},
{
"lang": "en",
"type": "remediation developer",
"value": "Claude Opus 5.5"
},
{
"lang": "en",
"type": "tool",
"value": "Claude Opus 5.5"
}
],
"cvssRationale": "AV:N: The feed key is in a URL and the API key is in a header; both are accessible over the network. AC:L: A single HTTP request with the token is sufficient; no race conditions or special timing required. AT:N: No manipulation of the target system is needed. PR:N: No prior authentication is required; the token itself is the credential, and it is not protected by any additional gate. UI:N: No user interaction is needed. VC/VI/VA:N: The vulnerable component (the authentication layer) is not itself compromised in terms of its own data, integrity, or availability. SC:H: The attacker gains full read access to the user\u0027s account data, including the API key and profile information. SI:H: The attacker can modify the user\u0027s account settings, and for administrator accounts, can perform arbitrary administrative actions via /admin. SA:N: No availability impact on the user\u0027s account or the application is demonstrated by the evidence.",
"draft": false,
"fixSummary": "The authentication decorators for feed-key and API-key requests no longer call Flask-Login\u0027s login_user(), which was writing the authenticated user into the session cookie and creating a full web session. Instead, the user object is now stored on Flask\u0027s request-scoped g._login_user attribute, which Flask-Login\u0027s current_user property reads first. This authenticates only the current request without setting any session cookie, preventing a limited-scope token from being elevated to a persistent, all-pages session that bypasses password and two-factor authentication.",
"generatedAt": "2026-11-16T08:23:33.762073Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 7,
"confidence": "high",
"model": "qwen3.8:27b",
"score": 6
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "36e0430891b60133c5817b45b93b89bb0cb70af762b9b90382b646f9c4ed5a33",
"patchSummary": "In website/decorators.py (auth_feedkey_func) and website/web/api/v1/common.py (auth_func), the import of login_user from flask_login is removed and replaced with an import of g from flask. The call login_user(user) (or login_user(user, remember=False)) is replaced with g._login_user = user. A new test file tests/test_token_auth_no_session.py is added with four test cases verifying that: (1) a feed-key request returns 200 but sets no session cookie, (2) an API-key request returns 200 but sets no session cookie, (3) a subsequent unauthenticated request to /user/profile is redirected (302), and (4) a 403 response from an API-key request also sets no session cookie.",
"patchTruncated": false,
"patches": [
{
"commit": "3605d5b65d340343cc4b612d40d454afed84302b",
"date": "Mon, 16 Nov 2026 08:56:35 +0100",
"patchSha256": "36e0430891b60133c5817b45b93b89bb0cb70af762b9b90382b646f9c4ed5a33",
"retrievedAt": "2026-11-16T08:22:31.738779Z",
"source": "https://github.com/vulnerability-lookup/vulnerability-lookup/commit/3605d5b65d340343cc4b612d40d454afed84302b.patch",
"sourceUrl": "https://github.com/vulnerability-lookup/vulnerability-lookup/commit/3605d5b65d340343cc4b612d40d454afed84302b.patch",
"subject": "fix: [security] A feed key or an API key no longer opens a"
}
],
"source": "https://github.com/vulnerability-lookup/vulnerability-lookup/commit/3605d5b65d340343cc4b612d40d454afed84302b.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "total"
}
],
"role": "Supplier",
"timestamp": "2026-11-16T08:23:33Z",
"version": "2.0.3"
},
"subject": "fix: [security] A feed key or an API key no longer opens a",
"weaknessRationale": [
{
"cweId": "CWE-287",
"rationale": "The authentication mechanism improperly elevated a limited-scope credential (feed key or API key intended for single-request authentication) into a full web session via login_user(), bypassing password and two-factor authentication. The authentication check succeeded but the session establishment was incorrect for the credential type."
},
{
"cweId": "CWE-613",
"rationale": "The commit message explicitly references CWE-613 as a related concern: sessions created before the fix remain valid until they expire and are not bound to the original credential. While the primary fix addresses the session creation, the lack of credential binding in existing sessions is a secondary weakness acknowledged by the developers."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"actor": "maintainer",
"description": "Vulnerability report received by e-mail from the reporter.",
"id": "evt-received-1",
"timestamp": "2026-11-15T16:36:10Z",
"type": "received"
},
{
"actor": "maintainer",
"description": "Report reproduced: a feed key or an API key opens a full web session, which also bypasses two-factor authentication.",
"id": "evt-validated-1",
"timestamp": "2026-11-16T07:40:02Z",
"type": "validated"
},
{
"actor": "maintainer",
"description": "Corrective change authored (3605d5b6): a feed key or an API key no longer opens a web session.",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/vulnerability-lookup/vulnerability-lookup/commit/3605d5b65d340343cc4b612d40d454afed84302b"
],
"timestamp": "2026-11-16T07:56:35Z",
"type": "fix-developed"
},
{
"actor": "maintainer",
"description": "Fix pushed to the public main branch.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/vulnerability-lookup/vulnerability-lookup/commit/3605d5b65d340343cc4b612d40d454afed84302b"
],
"timestamp": "2026-11-16T08:06:43Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20245"
}
]
}
},
"cveMetadata": {
"cveId": "CVE-2026-129703",
"datePublished": "2026-11-16T08:31:28.044959Z",
"dateReserved": "2026-11-16T08:36:44.944Z",
"dateUpdated": "2026-11-16T08:36:45.046151Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20245"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2.0"
}
GCVE-1-2026-20325 (CVE-2026-128637)
Vulnerability from gna-1 – Published: 2026-11-13 07:14 – Updated: 2026-11-13 07:15- CWE-601 - Open Redirect
| URL | Tags |
|---|---|
| https://github.com/MISP/misp-modules/commit/d115f… | patch |
| Vendor | Product | Version | |
|---|---|---|---|
| misp | misp-modules |
Affected:
0 , ≤ 3.1.4
(semver)
cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:* |
qwen3.8:27b
2 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-13 07:10 - Model
qwen3.8:27b- Input
-
https://github.com/MISP/misp-modules/commit/d115f00414e761b0b70c8c8671edac681de74a4c.patch
4b395e050af9… - Confidence
- high
| Commit | Subject | Patch SHA-256 |
|---|---|---|
d115f00414e7
|
fix: [website] restrict post-login redirects to local paths | 4b395e050af9… |
Fix summary
The fix introduces a _safe_next_url() validation function that restricts post-login redirect destinations to local paths or same-host HTTP(S) URLs. Any target that is empty, contains backslashes, control characters, leading/trailing whitespace, an unrecognized scheme, a mismatched host, or a protocol-relative form is rejected and the user is redirected to the home page (/). Accepted same-host URLs are normalized to local paths (scheme and authority stripped) while preserving query strings and fragments, ensuring the Location header always contains a relative path. This eliminates the ability to redirect users to arbitrary external destinations after authentication.
Patch summary
In website/app/account/account.py, the import of urlsplit and urlunsplit from urllib.parse is added. A new _safe_next_url() function is defined that validates the target string against a series of safety checks (whitespace, backslashes, control characters, scheme allowlist, host match, protocol-relative rejection, double-slash path rejection) and returns either a normalized local path or the fallback '/'. The login() handler's redirect call is changed from redirect(request.args.get('next') or '/') to redirect(_safe_next_url(request.args.get('next'))). A new test file website/tests/login_checks.py (172 lines) is added with comprehensive unit tests covering external rejection, malformed input rejection, local path preservation, same-host-with-port normalization, wrong-password non-redirect, CSRF enforcement, and remember-me cookie preservation. The website README is updated to document the new redirect validation behavior.
CVSS rationale
AV:N: exploitable over the network via a crafted URL. AC:L: no race conditions or special conditions; simply append a next parameter. AT:N: no attack requirements beyond delivering the URL. PR:N: the attacker needs no account; the victim performs the login. UI:A: the victim must actively visit the crafted URL and submit credentials. VC/VI/VA:N: the vulnerable system itself suffers no direct confidentiality, integrity, or availability loss. SC:L: subsequent confidentiality impact is low because the redirected user may be tricked into entering sensitive data on a phishing page, but the session cookie is domain-scoped and not directly exfiltrated. SI/SA:N: no subsequent integrity or availability impact is evidenced.
Weakness rationale
- CWE-601 The login endpoint accepted an unvalidated user-supplied URL in the next query parameter and used it directly as the redirect destination, allowing an attacker to redirect authenticated users to arbitrary external sites. This is a textbook open redirect (CWE-601).
Attack pattern rationale
- CAPEC-113 The patch directly addresses an open redirect: an attacker crafts a login URL with a malicious next parameter to redirect the victim to an external phishing site after authentication. CAPEC-113 describes exactly this attack pattern where a web application redirects to a user-controlled destination without proper validation. This is the closest and most precise CAPEC mapping available for this vulnerability class.
Assumptions to verify
- No specific affected or fixed version is identified in the patch metadata; the version range is recorded as unspecified.
- The CVSS UI metric is set to Active (A) on the assumption that the victim must click a crafted link and submit login credentials; if the login page is auto-submitted in a phishing context, UI could be argued as Passive, but the evidence supports Active.
- SC is set to Low based on the phishing/social-engineering risk of redirecting a user to an external site after login; no direct data exfiltration mechanism is present in the vulnerable code path.
- The CAPEC-113 mapping is the closest available pattern for open redirect; no more specific CAPEC exists for post-login redirect manipulation.
- The patch does not indicate whether this vulnerability was reported through a coordinated disclosure process; no finder or reporter credit is assigned beyond the remediation developer.
- The website component is assumed to be a Flask-based web application served over HTTP(S); the patch does not specify deployment topology or additional authentication layers.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
7 | 11 | high | 6 |
- 2026-11-13 05:49 UTC Fix developed Corrective change authored (d115f00414e761b0b70c8c8671edac681de74a4c): fix: [website] restrict post-login redirects to local paths https://github.com/MISP/misp-modules/commit/d115f00414e761b0b70c8c8671edac681de74a4c.patch
- 2026-11-13 07:10 UTC Fix available Patch retrieved from the supplied URL; retrieval time is used as the default availability time. https://github.com/MISP/misp-modules/commit/d115f00414e761b0b70c8c8671edac681de74a4c.patch
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:*"
],
"modules": [
"website"
],
"product": "misp-modules",
"programFiles": [
"website/app/account/account.py"
],
"repo": "https://github.com/MISP/misp-modules",
"vendor": "misp",
"versions": [
{
"lessThanOrEqual": "3.1.4",
"status": "affected",
"version": "0",
"versionType": "semver"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:*",
"versionEndIncluding": "3.1.4",
"versionStartIncluding": "0",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Alexandre Dulaunoy"
}
],
"descriptions": [
{
"lang": "en",
"value": "The MISP Modules website component contains an open redirect vulnerability in its login endpoint. The /login route accepted a user-supplied next query parameter and passed it directly to Flask\u0027s redirect() function without validating the destination. An attacker could craft a URL such as https://victim-host/login?next=https://attacker.example/phish and deliver it to a user (e.g., via email or a lookalike page). After the user successfully authenticated, the application issued a 302 redirect to the attacker-controlled external URL. This enables phishing attacks where a user, believing they are still on the legitimate MISP Modules site, may enter additional credentials or sensitive information on the attacker\u0027s page. It can also be used to obscure the true destination of a link, bypassing user trust in the legitimate domain. The vulnerability requires the victim to interact with the crafted URL and complete a login, but no prior authentication or special configuration is needed on the attacker\u0027s side. The affected component is the website sub-project of misp-modules, specifically the account blueprint in website/app/account/account.py. No version boundaries are specified in the patch metadata."
}
],
"impacts": [
{
"capecId": "CAPEC-113",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-113 Open Redirect"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "LOW",
"attackRequirements": "NONE",
"attackVector": "NETWORK",
"baseScore": 5.1,
"baseSeverity": "MEDIUM",
"privilegesRequired": "NONE",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "LOW",
"subIntegrityImpact": "NONE",
"userInteraction": "ACTIVE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:A/VC:N/VI:N/VA:N/SC:L/SI:N/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "NONE",
"vulnConfidentialityImpact": "NONE",
"vulnIntegrityImpact": "NONE",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-13T07:10:51Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-601",
"description": "CWE-601 Open Redirect",
"lang": "en",
"type": "CWE"
}
]
}
],
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/MISP/misp-modules/commit/d115f00414e761b0b70c8c8671edac681de74a4c"
}
],
"solutions": [
{
"lang": "en",
"value": "The fix introduces a _safe_next_url() validation function that restricts post-login redirect destinations to local paths or same-host HTTP(S) URLs. Any target that is empty, contains backslashes, control characters, leading/trailing whitespace, an unrecognized scheme, a mismatched host, or a protocol-relative form is rejected and the user is redirected to the home page (/). Accepted same-host URLs are normalized to local paths (scheme and authority stripped) while preserving query strings and fragments, ensuring the Location header always contains a relative path. This eliminates the ability to redirect users to arbitrary external destinations after authentication."
}
],
"title": "Open Redirect in MISP Modules Website Login via Unvalidated next Parameter",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "full",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"No specific affected or fixed version is identified in the patch metadata; the version range is recorded as unspecified.",
"The CVSS UI metric is set to Active (A) on the assumption that the victim must click a crafted link and submit login credentials; if the login page is auto-submitted in a phishing context, UI could be argued as Passive, but the evidence supports Active.",
"SC is set to Low based on the phishing/social-engineering risk of redirecting a user to an external site after login; no direct data exfiltration mechanism is present in the vulnerable code path.",
"The CAPEC-113 mapping is the closest available pattern for open redirect; no more specific CAPEC exists for post-login redirect manipulation.",
"The patch does not indicate whether this vulnerability was reported through a coordinated disclosure process; no finder or reporter credit is assigned beyond the remediation developer.",
"The website component is assumed to be a Flask-based web application served over HTTP(S); the patch does not specify deployment topology or additional authentication layers."
],
"capecRationale": [
{
"capecId": "CAPEC-113",
"rationale": "The patch directly addresses an open redirect: an attacker crafts a login URL with a malicious next parameter to redirect the victim to an external phishing site after authentication. CAPEC-113 describes exactly this attack pattern where a web application redirects to a user-controlled destination without proper validation. This is the closest and most precise CAPEC mapping available for this vulnerability class."
}
],
"commit": "d115f00414e761b0b70c8c8671edac681de74a4c",
"confidence": "high",
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Alexandre Dulaunoy"
}
],
"cvssRationale": "AV:N: exploitable over the network via a crafted URL. AC:L: no race conditions or special conditions; simply append a next parameter. AT:N: no attack requirements beyond delivering the URL. PR:N: the attacker needs no account; the victim performs the login. UI:A: the victim must actively visit the crafted URL and submit credentials. VC/VI/VA:N: the vulnerable system itself suffers no direct confidentiality, integrity, or availability loss. SC:L: subsequent confidentiality impact is low because the redirected user may be tricked into entering sensitive data on a phishing page, but the session cookie is domain-scoped and not directly exfiltrated. SI/SA:N: no subsequent integrity or availability impact is evidenced.",
"draft": false,
"fixSummary": "The fix introduces a _safe_next_url() validation function that restricts post-login redirect destinations to local paths or same-host HTTP(S) URLs. Any target that is empty, contains backslashes, control characters, leading/trailing whitespace, an unrecognized scheme, a mismatched host, or a protocol-relative form is rejected and the user is redirected to the home page (/). Accepted same-host URLs are normalized to local paths (scheme and authority stripped) while preserving query strings and fragments, ensuring the Location header always contains a relative path. This eliminates the ability to redirect users to arbitrary external destinations after authentication.",
"generatedAt": "2026-11-13T07:10:51.231558Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 6,
"confidence": "high",
"model": "qwen3.8:27b",
"score": 7
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "4b395e050af927c4d1a6033230f563b7d397ebcadcd72e6a56d7e61428aa31ea",
"patchSummary": "In website/app/account/account.py, the import of urlsplit and urlunsplit from urllib.parse is added. A new _safe_next_url() function is defined that validates the target string against a series of safety checks (whitespace, backslashes, control characters, scheme allowlist, host match, protocol-relative rejection, double-slash path rejection) and returns either a normalized local path or the fallback \u0027/\u0027. The login() handler\u0027s redirect call is changed from redirect(request.args.get(\u0027next\u0027) or \u0027/\u0027) to redirect(_safe_next_url(request.args.get(\u0027next\u0027))). A new test file website/tests/login_checks.py (172 lines) is added with comprehensive unit tests covering external rejection, malformed input rejection, local path preservation, same-host-with-port normalization, wrong-password non-redirect, CSRF enforcement, and remember-me cookie preservation. The website README is updated to document the new redirect validation behavior.",
"patchTruncated": false,
"patches": [
{
"commit": "d115f00414e761b0b70c8c8671edac681de74a4c",
"date": "Fri, 13 Nov 2026 06:49:26 +0100",
"patchSha256": "4b395e050af927c4d1a6033230f563b7d397ebcadcd72e6a56d7e61428aa31ea",
"retrievedAt": "2026-11-13T07:10:04.602453Z",
"source": "https://github.com/MISP/misp-modules/commit/d115f00414e761b0b70c8c8671edac681de74a4c.patch",
"sourceUrl": "https://github.com/MISP/misp-modules/commit/d115f00414e761b0b70c8c8671edac681de74a4c.patch",
"subject": "fix: [website] restrict post-login redirects to local paths"
}
],
"source": "https://github.com/MISP/misp-modules/commit/d115f00414e761b0b70c8c8671edac681de74a4c.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-13T07:10:51Z",
"version": "2.0.3"
},
"subject": "fix: [website] restrict post-login redirects to local paths",
"weaknessRationale": [
{
"cweId": "CWE-601",
"rationale": "The login endpoint accepted an unvalidated user-supplied URL in the next query parameter and used it directly as the redirect destination, allowing an attacker to redirect authenticated users to arbitrary external sites. This is a textbook open redirect (CWE-601)."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"description": "Corrective change authored (d115f00414e761b0b70c8c8671edac681de74a4c): fix: [website] restrict post-login redirects to local paths",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/MISP/misp-modules/commit/d115f00414e761b0b70c8c8671edac681de74a4c.patch"
],
"timestamp": "2026-11-13T05:49:26Z",
"type": "fix-developed"
},
{
"description": "Patch retrieved from the supplied URL; retrieval time is used as the default availability time.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/MISP/misp-modules/commit/d115f00414e761b0b70c8c8671edac681de74a4c.patch"
],
"timestamp": "2026-11-13T07:10:04.602453Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20325"
}
]
}
},
"cveMetadata": {
"cveId": "CVE-2026-128637",
"datePublished": "2026-11-13T07:14:10.164230Z",
"dateReserved": "2026-11-13T07:15:13.649Z",
"dateUpdated": "2026-11-13T07:15:13.746324Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20325"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2.0"
}
GCVE-1-2026-20141 (CVE-2026-127046)
Vulnerability from gna-1 – Published: 2026-11-12 12:01 – Updated: 2026-11-12 13:49| URL | Tags |
|---|---|
| https://github.com/MISP/misp-modules/commit/61c27… | patch |
| Vendor | Product | Version | |
|---|---|---|---|
| misp | misp-modules |
Affected:
0 , ≤ 3.1.4
(semver)
cpe:2.3:a:misp:misp_modules:*:*:*:*:*:*:*:* |
qwen3.8:27b
2 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-12 11:59 - Model
qwen3.8:27b- Input
-
https://github.com/MISP/misp-modules/commit/61c27893066cee80d9711649765ebc2bccd0e4af.patch
200a89910bce… - Confidence
- medium
| Commit | Subject | Patch SHA-256 |
|---|---|---|
61c27893066c
|
fix: [website] keep external tool API keys out of responses | 200a89910bce… |
Fix summary
The fix removes the api_key field from the ExternalTools.to_json() serializer so that /external_tools/list no longer returns credentials. The edit form is changed to a write-only pattern: the API key input starts empty, a blank or missing value on save retains the previously stored key, and a non-empty value replaces it. After a successful save the client clears the key field and does not propagate the credential into the in-memory tool list. Input validation is tightened to reject non-string or over-length key values. Server-side tool submissions continue to use the stored key from the database.
Patch summary
models.py: removed the api_key entry from ExternalTools.to_json(). external_tools.py (change_config endpoint): replaced the nested if-chain that required tool_api_key to be present and non-empty with a flat validation loop for tool_id/tool_name/tool_url, added a type-and-length check (string, max 60 chars) for tool_api_key when present, and made the field optional (write-only semantics). external_tools_core.py (change_config_core): changed tool.api_key assignment to be conditional on a truthy request_json.get('tool_api_key'). external_tools_index.html: initialized the form key field to empty string instead of tool.api_key, added placeholder and helper text, fixed a bug where the error-clearing selector referenced the wrong field, moved browser-state updates to after a successful HTTP response (res.ok check), and cleared the key input after save. README.md: documented the write-only key behavior. Tests: added browser-level (Playwright) and unit-level regression tests verifying the list endpoint omits api_key,空
CVSS rationale
AV:N: the /external_tools/list endpoint is reachable over the network via HTTP. AC:L: exploitation requires only a single GET request with no special conditions. AT:N: no attack prerequisites such as race conditions or specific timing. PR:N: regression tests confirm the endpoint returns 200 and previously included api_key for both authenticated and unauthenticated clients, indicating no privilege is required. UI:N: no victim interaction is needed; the attacker simply reads the response. VC:H: the exposed data is API keys (credentials) for external services, representing high confidentiality loss. VI:N and VA:N: no integrity or availability impact on the MISP Modules website. SC:H: the stolen API keys can be used to authenticate to external analysis/enrichment services, representing high confidentiality impact on a subsequent component. SI:N and SA:N: no integrity or availability impact on subsequent components is indicated.
Weakness rationale
- CWE-200 The /external_tools/list endpoint returned stored API keys in its JSON response to any client (authenticated or not), exposing credentials to unauthorized parties. The browser form also pre-populated the key in the DOM.
- CWE-522 API keys used to authenticate to external services were stored and transmitted without adequate protection: they appeared in API responses and in client-side form state, violating the principle that credentials should be write-only from the client perspective.
Attack pattern rationale
- CAPEC-126 The closest available CAPEC pattern. The attack involves an attacker obtaining credential material (API keys) from a web application response and using it to authenticate to a subsequent service. CAPEC-126 covers the theft of session/credential tokens from a web application. The mapping is imperfect because the keys are long-lived API credentials rather than short-lived session tokens, and the attack does not involve a session per se, but the underlying mechanism (reading a credential from an HTTP response and replaying it) is the same class of attack. No CAPEC pattern specifically covers 'sensitive data in API response' without a session-hijacking framing, so this is the best available match.
Assumptions to verify
- The /external_tools/list endpoint is accessible without authentication based on regression tests that verify the response for both authenticated and unauthenticated clients; however, the README notes that management operations require ADMIN_PASSWORD, so the list endpoint's access control may vary by deployment configuration.
- CVSS SC:H assumes the exposed API keys grant meaningful access to external services; the actual scope of access depends on the specific external tools configured and is not specified in the patch.
- No specific product version or release boundary is identified in the patch; the affected version range is therefore marked as unspecified.
- CAPEC-126 is the closest available pattern but is not a perfect match; the vulnerability is more precisely an API response information disclosure than a session hijacking. No CAPEC pattern specifically covers 'sensitive credential in API JSON response' without a session framing.
- The commit references issue #816 but no reporter or finder is named in the patch metadata or commit message.
- The 60-character maximum length validation for API keys is a new constraint introduced by the patch; it is unclear whether keys longer than 60 characters were previously valid, which could affect deployments with longer keys.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
6 | 11 | medium | 6 |
- 2026-11-12 08:37 UTC Fix developed Corrective change authored (61c27893066cee80d9711649765ebc2bccd0e4af): fix: [website] keep external tool API keys out of responses https://github.com/MISP/misp-modules/commit/61c27893066cee80d9711649765ebc2bccd0e4af.patch
- 2026-11-12 11:58 UTC Fix available Patch retrieved from the supplied URL; retrieval time is used as the default availability time. https://github.com/MISP/misp-modules/commit/61c27893066cee80d9711649765ebc2bccd0e4af.patch
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:misp:misp_modules:*:*:*:*:*:*:*:*"
],
"modules": [
"website/app/external_tools",
"website/app/models"
],
"product": "misp-modules",
"programFiles": [
"website/app/models/models.py",
"website/app/external_tools/external_tools.py",
"website/app/external_tools/external_tools_core.py",
"website/app/templates/external_tools/external_tools_index.html"
],
"repo": "https://github.com/MISP/misp-modules",
"vendor": "misp",
"versions": [
{
"lessThanOrEqual": "3.1.4",
"status": "affected",
"version": "0",
"versionType": "semver"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:*",
"versionEndIncluding": "3.1.4",
"versionStartIncluding": "0",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Alexandre Dulaunoy"
}
],
"descriptions": [
{
"lang": "en",
"value": "The MISP Modules website component contains an information disclosure vulnerability in the external tools management interface. The /external_tools/list endpoint returned stored API keys (the api_key field) in its JSON response via the ExternalTools model serializer. Additionally, the browser-based edit form pre-populated the API key input field with the stored credential value, exposing it in the DOM.\n\nThe /external_tools/list endpoint is accessible without authentication (the regression tests verify the response for both authenticated and unauthenticated clients), meaning any network-reachable client could retrieve the API keys for all configured external tools. These keys are used by the server to authenticate submissions to external analysis or enrichment services, so their disclosure allows an attacker to impersonate the MISP instance or directly access those external services.\n\nThe vulnerability also affected the edit workflow: the change_config endpoint previously required the API key to be present and non-empty on every save, forcing the key to round-trip through the client. The browser JavaScript stored the key in a reactive state object and updated the in-memory tool list with the credential value, further increasing the window of exposure in client-side memory and network traffic.\n\nAffected component: the website sub-application of MISP Modules, specifically the external_tools module (serializer in models.py, list endpoint, and the external_tools_index.html template). The impact is limited to confidentiality of the stored API keys; no integrity or availability impact on the MISP Modules website itself is indicated by the patch."
}
],
"impacts": [
{
"capecId": "CAPEC-126",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-126 Web Application Session Hijacking"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "LOW",
"attackRequirements": "NONE",
"attackVector": "NETWORK",
"baseScore": 9.2,
"baseSeverity": "CRITICAL",
"privilegesRequired": "NONE",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "HIGH",
"subIntegrityImpact": "NONE",
"userInteraction": "NONE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:H/SI:N/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "NONE",
"vulnConfidentialityImpact": "HIGH",
"vulnIntegrityImpact": "NONE",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-12T11:59:08Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-200",
"description": "CWE-200 Exposure of Sensitive Information to an Unauthorized Actor",
"lang": "en",
"type": "CWE"
}
]
},
{
"descriptions": [
{
"cweId": "CWE-522",
"description": "CWE-522 Insufficiently Protected Credentials",
"lang": "en",
"type": "CWE"
}
]
}
],
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/MISP/misp-modules/commit/61c27893066cee80d9711649765ebc2bccd0e4af"
}
],
"solutions": [
{
"lang": "en",
"value": "The fix removes the api_key field from the ExternalTools.to_json() serializer so that /external_tools/list no longer returns credentials. The edit form is changed to a write-only pattern: the API key input starts empty, a blank or missing value on save retains the previously stored key, and a non-empty value replaces it. After a successful save the client clears the key field and does not propagate the credential into the in-memory tool list. Input validation is tightened to reject non-string or over-length key values. Server-side tool submissions continue to use the stored key from the database."
}
],
"title": "MISP Modules Website Exposes External Tool API Keys in /external_tools/list Response",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "none",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"The /external_tools/list endpoint is accessible without authentication based on regression tests that verify the response for both authenticated and unauthenticated clients; however, the README notes that management operations require ADMIN_PASSWORD, so the list endpoint\u0027s access control may vary by deployment configuration.",
"CVSS SC:H assumes the exposed API keys grant meaningful access to external services; the actual scope of access depends on the specific external tools configured and is not specified in the patch.",
"No specific product version or release boundary is identified in the patch; the affected version range is therefore marked as unspecified.",
"CAPEC-126 is the closest available pattern but is not a perfect match; the vulnerability is more precisely an API response information disclosure than a session hijacking. No CAPEC pattern specifically covers \u0027sensitive credential in API JSON response\u0027 without a session framing.",
"The commit references issue #816 but no reporter or finder is named in the patch metadata or commit message.",
"The 60-character maximum length validation for API keys is a new constraint introduced by the patch; it is unclear whether keys longer than 60 characters were previously valid, which could affect deployments with longer keys."
],
"capecRationale": [
{
"capecId": "CAPEC-126",
"rationale": "The closest available CAPEC pattern. The attack involves an attacker obtaining credential material (API keys) from a web application response and using it to authenticate to a subsequent service. CAPEC-126 covers the theft of session/credential tokens from a web application. The mapping is imperfect because the keys are long-lived API credentials rather than short-lived session tokens, and the attack does not involve a session per se, but the underlying mechanism (reading a credential from an HTTP response and replaying it) is the same class of attack. No CAPEC pattern specifically covers \u0027sensitive data in API response\u0027 without a session-hijacking framing, so this is the best available match."
}
],
"commit": "61c27893066cee80d9711649765ebc2bccd0e4af",
"confidence": "medium",
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Alexandre Dulaunoy"
}
],
"cvssRationale": "AV:N: the /external_tools/list endpoint is reachable over the network via HTTP. AC:L: exploitation requires only a single GET request with no special conditions. AT:N: no attack prerequisites such as race conditions or specific timing. PR:N: regression tests confirm the endpoint returns 200 and previously included api_key for both authenticated and unauthenticated clients, indicating no privilege is required. UI:N: no victim interaction is needed; the attacker simply reads the response. VC:H: the exposed data is API keys (credentials) for external services, representing high confidentiality loss. VI:N and VA:N: no integrity or availability impact on the MISP Modules website. SC:H: the stolen API keys can be used to authenticate to external analysis/enrichment services, representing high confidentiality impact on a subsequent component. SI:N and SA:N: no integrity or availability impact on subsequent components is indicated.",
"draft": false,
"fixSummary": "The fix removes the api_key field from the ExternalTools.to_json() serializer so that /external_tools/list no longer returns credentials. The edit form is changed to a write-only pattern: the API key input starts empty, a blank or missing value on save retains the previously stored key, and a non-empty value replaces it. After a successful save the client clears the key field and does not propagate the credential into the in-memory tool list. Input validation is tightened to reject non-string or over-length key values. Server-side tool submissions continue to use the stored key from the database.",
"generatedAt": "2026-11-12T11:59:08.631868Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 6,
"confidence": "medium",
"model": "qwen3.8:27b",
"score": 6
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "200a89910bce4823c08ee129545edc9fe7177cb0de9c5555925aa075155ca61b",
"patchSummary": "models.py: removed the api_key entry from ExternalTools.to_json(). external_tools.py (change_config endpoint): replaced the nested if-chain that required tool_api_key to be present and non-empty with a flat validation loop for tool_id/tool_name/tool_url, added a type-and-length check (string, max 60 chars) for tool_api_key when present, and made the field optional (write-only semantics). external_tools_core.py (change_config_core): changed tool.api_key assignment to be conditional on a truthy request_json.get(\u0027tool_api_key\u0027). external_tools_index.html: initialized the form key field to empty string instead of tool.api_key, added placeholder and helper text, fixed a bug where the error-clearing selector referenced the wrong field, moved browser-state updates to after a successful HTTP response (res.ok check), and cleared the key input after save. README.md: documented the write-only key behavior. Tests: added browser-level (Playwright) and unit-level regression tests verifying the list endpoint omits api_key,\u7a7a",
"patchTruncated": false,
"patches": [
{
"commit": "61c27893066cee80d9711649765ebc2bccd0e4af",
"date": "Thu, 12 Nov 2026 09:37:45 +0100",
"patchSha256": "200a89910bce4823c08ee129545edc9fe7177cb0de9c5555925aa075155ca61b",
"retrievedAt": "2026-11-12T11:58:13.406720Z",
"source": "https://github.com/MISP/misp-modules/commit/61c27893066cee80d9711649765ebc2bccd0e4af.patch",
"sourceUrl": "https://github.com/MISP/misp-modules/commit/61c27893066cee80d9711649765ebc2bccd0e4af.patch",
"subject": "fix: [website] keep external tool API keys out of responses"
}
],
"source": "https://github.com/MISP/misp-modules/commit/61c27893066cee80d9711649765ebc2bccd0e4af.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-12T11:59:08Z",
"version": "2.0.3"
},
"subject": "fix: [website] keep external tool API keys out of responses",
"weaknessRationale": [
{
"cweId": "CWE-200",
"rationale": "The /external_tools/list endpoint returned stored API keys in its JSON response to any client (authenticated or not), exposing credentials to unauthorized parties. The browser form also pre-populated the key in the DOM."
},
{
"cweId": "CWE-522",
"rationale": "API keys used to authenticate to external services were stored and transmitted without adequate protection: they appeared in API responses and in client-side form state, violating the principle that credentials should be write-only from the client perspective."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"description": "Corrective change authored (61c27893066cee80d9711649765ebc2bccd0e4af): fix: [website] keep external tool API keys out of responses",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/MISP/misp-modules/commit/61c27893066cee80d9711649765ebc2bccd0e4af.patch"
],
"timestamp": "2026-11-12T08:37:45Z",
"type": "fix-developed"
},
{
"description": "Patch retrieved from the supplied URL; retrieval time is used as the default availability time.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/MISP/misp-modules/commit/61c27893066cee80d9711649765ebc2bccd0e4af.patch"
],
"timestamp": "2026-11-12T11:58:13.406720Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20141"
}
]
}
},
"cveMetadata": {
"cveId": "CVE-2026-127046",
"datePublished": "2026-11-12T12:01:50.481806Z",
"dateReserved": "2026-11-12T12:04:58.364Z",
"dateUpdated": "2026-11-12T13:49:40.957416Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20141"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2.0"
}
GCVE-1-2026-20323 (CVE-2026-127016)
Vulnerability from gna-1 – Published: 2026-11-12 09:40 – Updated: 2026-11-12 09:49- CWE-862 - Missing Authorization
| URL | Tags |
|---|---|
| https://github.com/MISP/misp-modules/commit/28c27… | patch |
| Vendor | Product | Version | |
|---|---|---|---|
| misp | misp-modules |
Affected:
0 , ≤ 3.1.4
(semver)
cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:* |
qwen3.8:27b
2 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-12 08:59 - Model
qwen3.8:27b- Input
-
https://github.com/MISP/misp-modules/commit/28c27e30f6d3a10e38fe9c54bb52f57ccf70d027.patch
a7fa238950e4… - Confidence
- medium
| Commit | Subject | Patch SHA-256 |
|---|---|---|
28c27e30f6d3
|
fix: [website] authorize external tool management | a7fa238950e4… |
Fix summary
A before_request authorization handler is added to the external_tools blueprint that checks whether the admin user is active (ADMIN_PASSWORD configured) and whether the current user is authenticated before any management endpoint executes. When both conditions indicate an unauthenticated request to a protected endpoint, the handler redirects to the login page (for the add endpoint) or returns a 403 Permission Denied response (for delete, status, and configuration endpoints). This enforces the same optional-admin policy already applied to module configuration, while preserving the existing open-management behavior when no admin password is configured.
Patch summary
The patch adds a before_request decorator on external_tools_blueprint in website/app/external_tools/external_tools.py that imports current_user from flask_login and admin_user_active from app.utils. The handler checks if the request endpoint is one of the four management endpoints (add_external_tool, delete_tool, change_status, change_config) and if the admin user is active but the current user is not authenticated, it either redirects to the login page or returns a 403 JSON response. The patch also adds a new test file website/tests/external_tools_checks.py with regression tests verifying that anonymous mutations are denied even with valid CSRF tokens, that authorization precedes JSON validation, that authenticated admins can perform all operations, and that the open-management policy is preserved when no admin password is set. The README is updated to document the new authorization requirement.
CVSS rationale
AV:N: The vulnerability is exploitable over the network via HTTP requests to the website. AC:L: No special conditions are required; standard HTTP POST/GET requests to known endpoints suffice. AT:N: No attack target manipulation is needed. PR:N: No privileges are required; the attack works as an anonymous user. UI:N: No user interaction is required. VC:L: Tool configurations including API keys and URLs are accessible/modifiable by the attacker. VI:H: The attacker can add, delete, and modify external tools, fully compromising the integrity of the tool management function. VA:L: Deleting or disabling tools reduces availability of data ingestion. SC:N: No secondary system confidentiality impact identified. SI:H: External tools feed into the broader MISP data pipeline; modifying them can corrupt or redirect data ingestion for the entire MISP instance. SA:N: No secondary availability impact beyond the direct effect.
Weakness rationale
- CWE-862 The external tools management endpoints (add, delete, change_status, change_config) lacked an authorization check that would verify the requester is an authenticated administrator when the admin-login policy is active. The authentication mechanism (flask_login, ADMIN_PASSWORD) existed in the application but was not applied to these specific endpoints, allowing anonymous users to perform privileged management operations.
Attack pattern rationale
- CAPEC-1 The closest available CAPEC pattern is Software Feature Discovery, as the attacker must identify the external tools management endpoints (e.g., /add_external_tool, /external_tools/<id>/delete_tool, /external_tools/change_status, /external_tools/change_config) and then issue unauthenticated requests to them. The vulnerability is exploited by discovering that these management functions are accessible without the authorization that the application's own documentation and parallel module-configuration endpoints imply. No CAPEC specifically models 'missing authorization on a discovered endpoint,' so CAPEC-1 is the best available match for the discovery-and-exploitation sequence.
Assumptions to verify
- The CAPEC mapping to CAPEC-1 (Software Feature Discovery) is the closest available pattern; no CAPEC specifically models exploitation of a missing authorization check on a discovered endpoint. The mapping is uncertain but is the best fit given the attack sequence of discovering and then invoking unauthenticated management endpoints.
- CVSS SI:H assumes that external tools in MISP feed into a broader data ingestion pipeline whose integrity is affected by unauthorized modification of tool configurations. If external tools are isolated to a single instance with no downstream consumers, SI could be reduced.
- The vulnerability is conditional on ADMIN_PASSWORD being configured. Installations without ADMIN_PASSWORD intentionally allow open management access by design and are not considered vulnerable.
- No specific version boundaries are identified in the patch; the affected version range is unspecified.
- The patch references issue #816 in the MISP/misp-modules repository, but no additional context about the issue tracker entry is available to confirm reporter or finder credits.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
7 | 11 | medium | 5 |
- 2026-11-12 08:37 UTC Fix developed Corrective change authored (28c27e30f6d3a10e38fe9c54bb52f57ccf70d027): fix: [website] authorize external tool management https://github.com/MISP/misp-modules/commit/28c27e30f6d3a10e38fe9c54bb52f57ccf70d027.patch
- 2026-11-12 08:58 UTC Fix available Patch retrieved from the supplied URL; retrieval time is used as the default availability time. https://github.com/MISP/misp-modules/commit/28c27e30f6d3a10e38fe9c54bb52f57ccf70d027.patch
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:*"
],
"modules": [
"external_tools"
],
"product": "misp-modules",
"programFiles": [
"website/app/external_tools/external_tools.py"
],
"repo": "https://github.com/MISP/misp-modules",
"vendor": "misp",
"versions": [
{
"lessThanOrEqual": "3.1.4",
"status": "affected",
"version": "0",
"versionType": "semver"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:*",
"versionEndIncluding": "3.1.4",
"versionStartIncluding": "0",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Alexandre Dulaunoy"
}
],
"descriptions": [
{
"lang": "en",
"value": "The MISP Modules website component exposes external tool management endpoints (add, delete, change status, and change configuration) that did not enforce the optional administrator authentication policy. When ADMIN_PASSWORD was configured in the application environment, the module configuration endpoints correctly required an authenticated administrator session, but the external tools management handlers processed requests from anonymous users without any authorization check. An unauthenticated attacker with network access to the website could add arbitrary external tools, delete existing tools, toggle their active status, and modify their configurations including API keys and ingestion URLs. This inconsistency allowed bypass of the intended access control for the external tools management functionality. The vulnerability is conditional on ADMIN_PASSWORD being set; installations without an admin password intentionally permit open management access by design. The affected component is the external_tools Flask blueprint within the MISP Modules website application."
}
],
"impacts": [
{
"capecId": "CAPEC-1",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-1 Software Feature Discovery"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "LOW",
"attackRequirements": "NONE",
"attackVector": "NETWORK",
"baseScore": 9.2,
"baseSeverity": "CRITICAL",
"privilegesRequired": "NONE",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "NONE",
"subIntegrityImpact": "HIGH",
"userInteraction": "NONE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:N/UI:N/VC:L/VI:H/VA:L/SC:N/SI:H/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "LOW",
"vulnConfidentialityImpact": "LOW",
"vulnIntegrityImpact": "HIGH",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "total"
}
],
"role": "Supplier",
"timestamp": "2026-11-12T08:59:10Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-862",
"description": "CWE-862 Missing Authorization",
"lang": "en",
"type": "CWE"
}
]
}
],
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/MISP/misp-modules/commit/28c27e30f6d3a10e38fe9c54bb52f57ccf70d027"
}
],
"solutions": [
{
"lang": "en",
"value": "A before_request authorization handler is added to the external_tools blueprint that checks whether the admin user is active (ADMIN_PASSWORD configured) and whether the current user is authenticated before any management endpoint executes. When both conditions indicate an unauthenticated request to a protected endpoint, the handler redirects to the login page (for the add endpoint) or returns a 403 Permission Denied response (for delete, status, and configuration endpoints). This enforces the same optional-admin policy already applied to module configuration, while preserving the existing open-management behavior when no admin password is configured."
}
],
"title": "MISP Modules Website: Missing Authorization on External Tool Management Endpoints",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "full",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"The CAPEC mapping to CAPEC-1 (Software Feature Discovery) is the closest available pattern; no CAPEC specifically models exploitation of a missing authorization check on a discovered endpoint. The mapping is uncertain but is the best fit given the attack sequence of discovering and then invoking unauthenticated management endpoints.",
"CVSS SI:H assumes that external tools in MISP feed into a broader data ingestion pipeline whose integrity is affected by unauthorized modification of tool configurations. If external tools are isolated to a single instance with no downstream consumers, SI could be reduced.",
"The vulnerability is conditional on ADMIN_PASSWORD being configured. Installations without ADMIN_PASSWORD intentionally allow open management access by design and are not considered vulnerable.",
"No specific version boundaries are identified in the patch; the affected version range is unspecified.",
"The patch references issue #816 in the MISP/misp-modules repository, but no additional context about the issue tracker entry is available to confirm reporter or finder credits."
],
"capecRationale": [
{
"capecId": "CAPEC-1",
"rationale": "The closest available CAPEC pattern is Software Feature Discovery, as the attacker must identify the external tools management endpoints (e.g., /add_external_tool, /external_tools/\u003cid\u003e/delete_tool, /external_tools/change_status, /external_tools/change_config) and then issue unauthenticated requests to them. The vulnerability is exploited by discovering that these management functions are accessible without the authorization that the application\u0027s own documentation and parallel module-configuration endpoints imply. No CAPEC specifically models \u0027missing authorization on a discovered endpoint,\u0027 so CAPEC-1 is the best available match for the discovery-and-exploitation sequence."
}
],
"commit": "28c27e30f6d3a10e38fe9c54bb52f57ccf70d027",
"confidence": "medium",
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Alexandre Dulaunoy"
}
],
"cvssRationale": "AV:N: The vulnerability is exploitable over the network via HTTP requests to the website. AC:L: No special conditions are required; standard HTTP POST/GET requests to known endpoints suffice. AT:N: No attack target manipulation is needed. PR:N: No privileges are required; the attack works as an anonymous user. UI:N: No user interaction is required. VC:L: Tool configurations including API keys and URLs are accessible/modifiable by the attacker. VI:H: The attacker can add, delete, and modify external tools, fully compromising the integrity of the tool management function. VA:L: Deleting or disabling tools reduces availability of data ingestion. SC:N: No secondary system confidentiality impact identified. SI:H: External tools feed into the broader MISP data pipeline; modifying them can corrupt or redirect data ingestion for the entire MISP instance. SA:N: No secondary availability impact beyond the direct effect.",
"draft": false,
"fixSummary": "A before_request authorization handler is added to the external_tools blueprint that checks whether the admin user is active (ADMIN_PASSWORD configured) and whether the current user is authenticated before any management endpoint executes. When both conditions indicate an unauthenticated request to a protected endpoint, the handler redirects to the login page (for the add endpoint) or returns a 403 Permission Denied response (for delete, status, and configuration endpoints). This enforces the same optional-admin policy already applied to module configuration, while preserving the existing open-management behavior when no admin password is configured.",
"generatedAt": "2026-11-12T08:59:10.622852Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 5,
"confidence": "medium",
"model": "qwen3.8:27b",
"score": 7
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "a7fa238950e4db922c8ad604a6abe453385d29b8a6bcc58762a4fe72add73775",
"patchSummary": "The patch adds a before_request decorator on external_tools_blueprint in website/app/external_tools/external_tools.py that imports current_user from flask_login and admin_user_active from app.utils. The handler checks if the request endpoint is one of the four management endpoints (add_external_tool, delete_tool, change_status, change_config) and if the admin user is active but the current user is not authenticated, it either redirects to the login page or returns a 403 JSON response. The patch also adds a new test file website/tests/external_tools_checks.py with regression tests verifying that anonymous mutations are denied even with valid CSRF tokens, that authorization precedes JSON validation, that authenticated admins can perform all operations, and that the open-management policy is preserved when no admin password is set. The README is updated to document the new authorization requirement.",
"patchTruncated": false,
"patches": [
{
"commit": "28c27e30f6d3a10e38fe9c54bb52f57ccf70d027",
"date": "Thu, 12 Nov 2026 09:37:23 +0100",
"patchSha256": "a7fa238950e4db922c8ad604a6abe453385d29b8a6bcc58762a4fe72add73775",
"retrievedAt": "2026-11-12T08:58:21.777259Z",
"source": "https://github.com/MISP/misp-modules/commit/28c27e30f6d3a10e38fe9c54bb52f57ccf70d027.patch",
"sourceUrl": "https://github.com/MISP/misp-modules/commit/28c27e30f6d3a10e38fe9c54bb52f57ccf70d027.patch",
"subject": "fix: [website] authorize external tool management"
}
],
"source": "https://github.com/MISP/misp-modules/commit/28c27e30f6d3a10e38fe9c54bb52f57ccf70d027.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "total"
}
],
"role": "Supplier",
"timestamp": "2026-11-12T08:59:10Z",
"version": "2.0.3"
},
"subject": "fix: [website] authorize external tool management",
"weaknessRationale": [
{
"cweId": "CWE-862",
"rationale": "The external tools management endpoints (add, delete, change_status, change_config) lacked an authorization check that would verify the requester is an authenticated administrator when the admin-login policy is active. The authentication mechanism (flask_login, ADMIN_PASSWORD) existed in the application but was not applied to these specific endpoints, allowing anonymous users to perform privileged management operations."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"description": "Corrective change authored (28c27e30f6d3a10e38fe9c54bb52f57ccf70d027): fix: [website] authorize external tool management",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/MISP/misp-modules/commit/28c27e30f6d3a10e38fe9c54bb52f57ccf70d027.patch"
],
"timestamp": "2026-11-12T08:37:23Z",
"type": "fix-developed"
},
{
"description": "Patch retrieved from the supplied URL; retrieval time is used as the default availability time.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/MISP/misp-modules/commit/28c27e30f6d3a10e38fe9c54bb52f57ccf70d027.patch"
],
"timestamp": "2026-11-12T08:58:21.777259Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20323"
}
]
}
},
"cveMetadata": {
"cveId": "CVE-2026-127016",
"datePublished": "2026-11-12T09:40:23.014021Z",
"dateReserved": "2026-11-12T09:45:10.065Z",
"dateUpdated": "2026-11-12T09:49:37.028410Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20323"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2.0"
}
GCVE-1-2026-20333 (CVE-2026-126991)
Vulnerability from gna-1 – Published: 2026-11-12 07:28 – Updated: 2026-11-12 07:33| Vendor | Product | Version | |
|---|---|---|---|
| misp | misp-modules |
Affected:
unspecified , ≤ 3.1.4
(semver)
cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:* |
qwen3.8:27b
2 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-12 05:54 - Model
qwen3.8:27b- Input
-
https://github.com/MISP/misp-modules/commit/7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9.patch
5b0d882d17ba… - Confidence
- medium
| Commit | Subject | Patch SHA-256 |
|---|---|---|
7a2a7e3d9e0f
|
Improve Wikidata enrichment and return MISP standard | 5b0d882d17ba… |
Fix summary
The vulnerability is remediated by properly escaping user-supplied text as a SPARQL string literal using json.dumps with ensure_ascii=False, which neutralizes quotes, backslashes, newlines, and control characters. Additional hardening includes strict validation of the language tag, numeric configuration values, text length, and UUID format before query construction; validation of Wikidata response structure and entity URI format before importing results; replacement of raw exception formatting with generic error messages that do not expose upstream details; use of the modern Wikidata SPARQL endpoint with POST method; a proper MISP user-agent string; and a bounded request timeout.
Patch summary
The patch rewrites the wiki.py handler to: (1) escape user text via json.dumps(text, ensure_ascii=False) before embedding in the SPARQL query; (2) validate language tags with a regex, numeric settings (max_results 1-25, timeout 1-60), text length (1-2000 chars), and UUID format; (3) switch from string concatenation to a properly escaped SPARQL literal; (4) use the modern https://query.wikidata.org/sparql endpoint with POST method and a MISP user-agent; (5) validate Wikidata response bindings and entity URIs against a strict pattern before creating annotation objects; (6) replace the mutable shared misperrors dict and raw exception formatting with a ModuleError class and generic error strings; (7) add a 313-line test suite (tests/test_wiki.py) covering injection vectors, invalid inputs, malformed responses, and error sanitization; (8) update documentation and module metadata to reflect the new misp_standard output format and configuration options.
CVSS rationale
AV:N: The module is invoked over the network via the MISP API. AC:L: The injection requires only crafting a text value with SPARQL metacharacters; no race conditions or complex prerequisites. AT:N: No need to manipulate the attack target. PR:L: Requires authenticated access to a MISP instance with the wiki module enabled. UI:N: No user interaction beyond the API call. VC:N: The queried data is public Wikidata; no confidential MISP data is exposed. VI:L: An attacker can alter the enrichment results (integrity of advisory annotations) and the old error path could leak internal details. VA:N: No meaningful availability impact on the MISP system. SC/SI/SA:N: No impact on external systems beyond the public Wikidata service. Overall severity is low given the limited blast radius.
Weakness rationale
- CWE-94 The module dynamically constructs a SPARQL query by concatenating untrusted user input directly into the query string without neutralizing special elements. This is the same pattern as code/query injection: the intended query structure is modified by attacker-controlled input. SPARQL is a query language interpreted by the Wikidata endpoint, making this a code injection into a query interpreter.
- CWE-209 The old code formatted raw exception objects into the error response string ('wikidata API not accessible {}'.format(e)), which could expose internal details such as proxy credentials, full query text, or upstream response bodies. The fix replaces this with generic error messages.
Attack pattern rationale
- CAPEC-143 CAPEC-143 describes injection of query language fragments through unsanitized input into a dynamically constructed query. Although the CAPEC name references SQL specifically, the attack pattern is identical: an attacker supplies input containing query-language metacharacters (quotes, braces, keywords) that alter the structure of the query sent to an interpreter. SPARQL injection follows the same pattern. This is the closest available CAPEC; no SPARQL-specific CAPEC exists in the catalog. The uncertainty is that the CAPEC is SQL-specific in naming, but the underlying technique (query injection via string concatenation) is the same.
Assumptions to verify
- The CAPEC-143 mapping is the closest available pattern; no SPARQL-specific CAPEC exists in the catalog. The underlying technique (query injection via unsanitized string concatenation) is identical to SQL injection.
- CVSS PR:L assumes the attacker has authenticated access to a MISP instance with the wiki expansion module enabled. The actual privilege level required depends on MISP role configuration, which is not specified in the patch.
- The affected version range is unspecified; the patch does not include version tags or release boundaries. The vulnerability existed in module version 0.2 (as indicated by the old moduleinfo) and was fixed in version 1.0.
- The CVSS VI:L rating accounts for both the query manipulation and the old error-message information disclosure. If only the injection is considered (without the error leak), VI could be rated N.
- The patch is treated as a single vulnerability fix (SPARQL injection) with associated hardening changes (error sanitization, input validation, endpoint update). The caller asserts all changes relate to the same vulnerability.
- No evidence of active exploitation, PoC, or in-the-wild usage was found in the patch or metadata.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
6 | 11 | medium | 6 |
- 2026-11-12 04:47 UTC Fix developed Corrective change authored (7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9): Improve Wikidata enrichment and return MISP standard https://github.com/MISP/misp-modules/commit/7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9.patch
- 2026-11-12 05:53 UTC Fix available Patch retrieved from the supplied URL; retrieval time is used as the default availability time. https://github.com/MISP/misp-modules/commit/7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9.patch
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:*"
],
"modules": [
"misp_modules/modules/expansion/wiki.py"
],
"product": "misp-modules",
"programFiles": [
"misp_modules/modules/expansion/wiki.py"
],
"repo": "https://github.com/MISP/misp-modules",
"vendor": "misp",
"versions": [
{
"lessThanOrEqual": "3.1.4",
"status": "affected",
"version": "unspecified",
"versionType": "semver"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:*",
"versionEndExcluding": "3.1.4",
"versionStartIncluding": "0",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Alexandre Dulaunoy"
}
],
"descriptions": [
{
"lang": "en",
"value": "The Wikidata Lookup expansion module (wiki.py) in MISP misp-modules contained a SPARQL injection vulnerability. The handler function constructed a SPARQL query by directly concatenating the user-supplied text value into the query string without any escaping or parameterization: \u0027SELECT ?item WHERE { ?item rdfs:label\"\u0027 + request.get(\"text\") + \u0027\" @en }\u0027. An authenticated MISP user with access to the module could supply a crafted text attribute value containing SPARQL syntax (quotes, closing braces, UNION, SERVICE, LIMIT, etc.) to alter the structure of the query sent to the Wikidata SPARQL endpoint. This could allow an attacker to modify the query logic, extract unintended data, or cause the query to fail in ways that expose internal error details. Additionally, the old error handling path formatted raw exception objects into the response string (\u0027wikidata API not accessible {}\u0027.format(e)), potentially leaking internal information such as proxy credentials, query contents, or response bodies. The module also used a spoofed browser user-agent string and the deprecated Wikidata endpoint. The vulnerability required authenticated access to the MISP instance and the ability to invoke the wiki expansion module with a text attribute. The impact is limited to manipulation of the enrichment query and potential disclosure of internal error details; it does not directly compromise the MISP server or its stored data."
}
],
"impacts": [
{
"capecId": "CAPEC-143",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-143 SQL Injection"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "LOW",
"attackRequirements": "NONE",
"attackVector": "NETWORK",
"baseScore": 5.3,
"baseSeverity": "MEDIUM",
"privilegesRequired": "LOW",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "NONE",
"subIntegrityImpact": "NONE",
"userInteraction": "NONE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:L/VA:N/SC:N/SI:N/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "NONE",
"vulnConfidentialityImpact": "NONE",
"vulnIntegrityImpact": "LOW",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-12T05:54:11Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-94",
"description": "CWE-94 Improper Control of Generation of Code (\u0027Code Injection\u0027)",
"lang": "en",
"type": "CWE"
}
]
},
{
"descriptions": [
{
"cweId": "CWE-209",
"description": "CWE-209 Generation of Error Message Containing Sensitive Information",
"lang": "en",
"type": "CWE"
}
]
}
],
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/MISP/misp-modules/commit/7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9"
},
{
"url": "https://misp.github.io/misp-modules/expansion/#wikidata-lookup"
},
{
"url": "https://github.com/MISP/misp-modules/pull/869"
},
{
"url": "https://github.com/MISP/misp-modules"
}
],
"solutions": [
{
"lang": "en",
"value": "The vulnerability is remediated by properly escaping user-supplied text as a SPARQL string literal using json.dumps with ensure_ascii=False, which neutralizes quotes, backslashes, newlines, and control characters. Additional hardening includes strict validation of the language tag, numeric configuration values, text length, and UUID format before query construction; validation of Wikidata response structure and entity URI format before importing results; replacement of raw exception formatting with generic error messages that do not expose upstream details; use of the modern Wikidata SPARQL endpoint with POST method; a proper MISP user-agent string; and a bounded request timeout."
}
],
"title": "SPARQL Injection in MISP Wikidata Lookup Module via Unescaped User Input",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "partial",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"The CAPEC-143 mapping is the closest available pattern; no SPARQL-specific CAPEC exists in the catalog. The underlying technique (query injection via unsanitized string concatenation) is identical to SQL injection.",
"CVSS PR:L assumes the attacker has authenticated access to a MISP instance with the wiki expansion module enabled. The actual privilege level required depends on MISP role configuration, which is not specified in the patch.",
"The affected version range is unspecified; the patch does not include version tags or release boundaries. The vulnerability existed in module version 0.2 (as indicated by the old moduleinfo) and was fixed in version 1.0.",
"The CVSS VI:L rating accounts for both the query manipulation and the old error-message information disclosure. If only the injection is considered (without the error leak), VI could be rated N.",
"The patch is treated as a single vulnerability fix (SPARQL injection) with associated hardening changes (error sanitization, input validation, endpoint update). The caller asserts all changes relate to the same vulnerability.",
"No evidence of active exploitation, PoC, or in-the-wild usage was found in the patch or metadata."
],
"capecRationale": [
{
"capecId": "CAPEC-143",
"rationale": "CAPEC-143 describes injection of query language fragments through unsanitized input into a dynamically constructed query. Although the CAPEC name references SQL specifically, the attack pattern is identical: an attacker supplies input containing query-language metacharacters (quotes, braces, keywords) that alter the structure of the query sent to an interpreter. SPARQL injection follows the same pattern. This is the closest available CAPEC; no SPARQL-specific CAPEC exists in the catalog. The uncertainty is that the CAPEC is SQL-specific in naming, but the underlying technique (query injection via string concatenation) is the same."
}
],
"commit": "7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9",
"confidence": "medium",
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "Alexandre Dulaunoy"
}
],
"cvssRationale": "AV:N: The module is invoked over the network via the MISP API. AC:L: The injection requires only crafting a text value with SPARQL metacharacters; no race conditions or complex prerequisites. AT:N: No need to manipulate the attack target. PR:L: Requires authenticated access to a MISP instance with the wiki module enabled. UI:N: No user interaction beyond the API call. VC:N: The queried data is public Wikidata; no confidential MISP data is exposed. VI:L: An attacker can alter the enrichment results (integrity of advisory annotations) and the old error path could leak internal details. VA:N: No meaningful availability impact on the MISP system. SC/SI/SA:N: No impact on external systems beyond the public Wikidata service. Overall severity is low given the limited blast radius.",
"draft": false,
"fixSummary": "The vulnerability is remediated by properly escaping user-supplied text as a SPARQL string literal using json.dumps with ensure_ascii=False, which neutralizes quotes, backslashes, newlines, and control characters. Additional hardening includes strict validation of the language tag, numeric configuration values, text length, and UUID format before query construction; validation of Wikidata response structure and entity URI format before importing results; replacement of raw exception formatting with generic error messages that do not expose upstream details; use of the modern Wikidata SPARQL endpoint with POST method; a proper MISP user-agent string; and a bounded request timeout.",
"generatedAt": "2026-11-12T05:54:11.233388Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 6,
"confidence": "medium",
"model": "qwen3.8:27b",
"score": 6
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "5b0d882d17baea052254770c3c94d50c8e1ab2314fa4eec89088c15e059f2837",
"patchSummary": "The patch rewrites the wiki.py handler to: (1) escape user text via json.dumps(text, ensure_ascii=False) before embedding in the SPARQL query; (2) validate language tags with a regex, numeric settings (max_results 1-25, timeout 1-60), text length (1-2000 chars), and UUID format; (3) switch from string concatenation to a properly escaped SPARQL literal; (4) use the modern https://query.wikidata.org/sparql endpoint with POST method and a MISP user-agent; (5) validate Wikidata response bindings and entity URIs against a strict pattern before creating annotation objects; (6) replace the mutable shared misperrors dict and raw exception formatting with a ModuleError class and generic error strings; (7) add a 313-line test suite (tests/test_wiki.py) covering injection vectors, invalid inputs, malformed responses, and error sanitization; (8) update documentation and module metadata to reflect the new misp_standard output format and configuration options.",
"patchTruncated": false,
"patches": [
{
"commit": "7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9",
"date": "Thu, 12 Nov 2026 04:47:41 +0000",
"patchSha256": "5b0d882d17baea052254770c3c94d50c8e1ab2314fa4eec89088c15e059f2837",
"retrievedAt": "2026-11-12T05:53:20.752027Z",
"source": "https://github.com/MISP/misp-modules/commit/7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9.patch",
"sourceUrl": "https://github.com/MISP/misp-modules/commit/7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9.patch",
"subject": "Improve Wikidata enrichment and return MISP standard"
}
],
"source": "https://github.com/MISP/misp-modules/commit/7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-12T05:54:11Z",
"version": "2.0.3"
},
"subject": "Improve Wikidata enrichment and return MISP standard",
"weaknessRationale": [
{
"cweId": "CWE-94",
"rationale": "The module dynamically constructs a SPARQL query by concatenating untrusted user input directly into the query string without neutralizing special elements. This is the same pattern as code/query injection: the intended query structure is modified by attacker-controlled input. SPARQL is a query language interpreted by the Wikidata endpoint, making this a code injection into a query interpreter."
},
{
"cweId": "CWE-209",
"rationale": "The old code formatted raw exception objects into the error response string (\u0027wikidata API not accessible {}\u0027.format(e)), which could expose internal details such as proxy credentials, full query text, or upstream response bodies. The fix replaces this with generic error messages."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"description": "Corrective change authored (7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9): Improve Wikidata enrichment and return MISP standard",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/MISP/misp-modules/commit/7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9.patch"
],
"timestamp": "2026-11-12T04:47:41Z",
"type": "fix-developed"
},
{
"description": "Patch retrieved from the supplied URL; retrieval time is used as the default availability time.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/MISP/misp-modules/commit/7a2a7e3d9e0f67f6732bd8050ab7da79e7b344d9.patch"
],
"timestamp": "2026-11-12T05:53:20.752027Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20333"
}
]
}
},
"cveMetadata": {
"cveId": "CVE-2026-126991",
"datePublished": "2026-11-12T07:28:09.005889Z",
"dateReserved": "2026-11-12T07:30:41.366Z",
"dateUpdated": "2026-11-12T07:33:42.155982Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20333"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2.0"
}
GCVE-1-2026-20273 (CVE-2026-125871)
Vulnerability from gna-1 – Published: 2026-11-10 12:48 – Updated: 2026-11-10 15:01| URL | Tags |
|---|---|
| https://github.com/MISP/misp-modules/commit/7cb31… | patch |
| Vendor | Product | Version | |
|---|---|---|---|
| MISP | misp-modules |
Affected:
0 , < 3.0.13
(semver)
cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:* |
qwen3.8:27b
2 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-10 11:02 - Model
qwen3.8:27b- Input
-
https://github.com/MISP/misp-modules/commit/7cb31e96a1f59bec9afbc087b27b26d9986c6841.patch
736974194693… - Confidence
- high
| Commit | Subject | Patch SHA-256 |
|---|---|---|
7cb31e96a1f5
|
fix: [sourcecache] validate URLs before archiving to block | 736974194693… |
Fix summary
The fix introduces a URL validation gate (is_safe_url) that is invoked before any archiving request is made. The function restricts the scheme to http or https, resolves the hostname via DNS, and verifies that every resolved IP address is globally routable (i.e., not loopback, private, link-local, or otherwise non-global). If any check fails, the module returns an error and the fetch is never performed, preventing the server from being directed to internal or non-HTTP targets.
Patch summary
Adds imports for ipaddress and socket. Introduces a new is_safe_url(url) function that parses the URL, rejects non-http/https schemes, resolves the hostname with socket.getaddrinfo, and rejects the URL if any resolved address fails ipaddress.ip_address().is_global(). In the handler() function, a guard clause calls is_safe_url(tocache) before invoking __archiveLink and returns an error dict if the URL is not safe. Two new unit tests verify that http://127.0.0.1:6379/ and file:///etc/passwd are blocked and that the archive function is never called.
CVSS rationale
AV:N: The MISP API is network-accessible. AC:L: Crafting a URL attribute with an internal address is trivial. AT:N: No special adversary technique is required. PR:L: An authenticated MISP user with attribute-creation rights is needed. UI:N: No victim interaction beyond the attacker's own API call. VC/VI/VA:N: The MISP server's own confidentiality, integrity, and availability are not directly compromised. SC:H: The subsequent component (internal services, cloud metadata) suffers high confidentiality impact because its data is exfiltrated into the MISP event. SI:N: The module only performs GET-style fetches; no evidence of integrity impact on subsequent components. SA:N: No availability impact on subsequent components is indicated.
Weakness rationale
- CWE-918 The module accepts a user-supplied URL and issues a server-side HTTP request to it without validating the target, allowing an attacker to direct requests to internal or non-HTTP resources. This is a textbook SSRF.
- CWE-20 The root cause is the absence of any validation on the URL attribute value before it is used as a request target. The fix adds scheme, hostname, and IP-address validation, confirming the missing input validation was the defect.
Attack pattern rationale
- CAPEC-12 The patch explicitly describes and remediates an SSRF: an attacker-supplied URL causes the MISP server to fetch internal resources (loopback, cloud metadata) and return the data. CAPEC-12 is the direct and most specific attack pattern for this scenario. No uncertainty is noted; the commit message and code changes align precisely with this pattern.
Assumptions to verify
- The MISP server is assumed to be reachable over the network and that an authenticated user with attribute-creation permissions can submit link/url attributes; no specific MISP version or authentication model is stated in the patch.
- CVSS PR:L assumes a low-privilege authenticated MISP user can create attributes; if MISP requires higher-level permissions, PR could be H.
- CVSS SI:N assumes the archiving function only performs read (GET) requests; if it can issue state-changing methods, SI could be elevated.
- SSVC technical_impact 'partial' is the least-assumptive choice; if cloud metadata endpoint access leads to full credential theft and infrastructure takeover, impact could be 'total'.
- The CAPEC-12 mapping is high-confidence based on the explicit SSRF description in the commit message and the code changes; no uncertainty is expected.
- No specific affected or fixed version numbers are provided in the patch metadata; version boundaries are marked as unspecified.
- The is_global check in Python's ipaddress module covers loopback, private, link-local, and other non-global ranges, but DNS rebinding between the check and the actual fetch is not addressed by this patch and could remain a residual risk.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
6 | 11 | high | 7 |
- 2026-10-16 22:27 UTC Fix developed Corrective change authored (7cb31e96a1f59bec9afbc087b27b26d9986c6841): fix: [sourcecache] validate URLs before archiving to block https://github.com/MISP/misp-modules/commit/7cb31e96a1f59bec9afbc087b27b26d9986c6841.patch
- 2026-11-10 11:01 UTC Fix available Patch retrieved from the supplied URL; retrieval time is used as the default availability time. https://github.com/MISP/misp-modules/commit/7cb31e96a1f59bec9afbc087b27b26d9986c6841.patch
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:misp:misp-modules:*:*:*:*:*:*:*:*"
],
"modules": [
"sourcecache"
],
"product": "misp-modules",
"programFiles": [
"misp_modules/modules/expansion/sourcecache.py"
],
"repo": "https://github.com/MISP/misp-modules",
"vendor": "MISP",
"versions": [
{
"lessThan": "3.0.13",
"status": "affected",
"version": "0",
"versionType": "semver"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:misp:misp_modules:*:*:*:*:*:*:*:*",
"versionEndExcluding": "3.0.13",
"versionStartIncluding": "0",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "elhoim"
},
{
"lang": "en",
"type": "remediation developer",
"value": "Claude Opus 5 (1M context)"
}
],
"descriptions": [
{
"lang": "en",
"value": "The sourcecache expansion module in MISP Modules (misp-modules) contains a Server-Side Request Forgery (SSRF) vulnerability. When a user with permission to create or modify attributes of type \u0027link\u0027 or \u0027url\u0027 submits an attribute value, the module passes the URL directly to its internal archiving function without any validation of the target address. An attacker can therefore craft a URL pointing to loopback addresses (e.g., 127.0.0.1), private network ranges, link-local addresses, or cloud metadata endpoints (e.g., 169.254.169.254). The MISP server will issue an HTTP request to that target and store the response body in the event, effectively exfiltrating data from internal services to the attacker through the MISP interface. Non-HTTP schemes such as file:// are also accepted, potentially allowing local file disclosure. The vulnerability requires an authenticated MISP user with the ability to create or edit attributes of the affected types. The security impact is information disclosure from internal services reachable by the MISP server, including sensitive data such as cloud credentials, internal application data, or local files."
}
],
"impacts": [
{
"capecId": "CAPEC-12",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-12 Exploiting SSRF"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "LOW",
"attackRequirements": "NONE",
"attackVector": "NETWORK",
"baseScore": 6.3,
"baseSeverity": "MEDIUM",
"privilegesRequired": "LOW",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "HIGH",
"subIntegrityImpact": "NONE",
"userInteraction": "NONE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:N/AC:L/AT:N/PR:L/UI:N/VC:N/VI:N/VA:N/SC:H/SI:N/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "NONE",
"vulnConfidentialityImpact": "NONE",
"vulnIntegrityImpact": "NONE",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-10T11:02:25Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-918",
"description": "CWE-918 Server-Side Request Forgery (SSRF)",
"lang": "en",
"type": "CWE"
}
]
},
{
"descriptions": [
{
"cweId": "CWE-20",
"description": "CWE-20 Improper Input Validation",
"lang": "en",
"type": "CWE"
}
]
}
],
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/MISP/misp-modules/commit/7cb31e96a1f59bec9afbc087b27b26d9986c6841"
}
],
"solutions": [
{
"lang": "en",
"value": "The fix introduces a URL validation gate (is_safe_url) that is invoked before any archiving request is made. The function restricts the scheme to http or https, resolves the hostname via DNS, and verifies that every resolved IP address is globally routable (i.e., not loopback, private, link-local, or otherwise non-global). If any check fails, the module returns an error and the fetch is never performed, preventing the server from being directed to internal or non-HTTP targets."
}
],
"title": "MISP sourcecache module SSRF allows fetching arbitrary internal URLs via attribute enrichment",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "full",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"The MISP server is assumed to be reachable over the network and that an authenticated user with attribute-creation permissions can submit link/url attributes; no specific MISP version or authentication model is stated in the patch.",
"CVSS PR:L assumes a low-privilege authenticated MISP user can create attributes; if MISP requires higher-level permissions, PR could be H.",
"CVSS SI:N assumes the archiving function only performs read (GET) requests; if it can issue state-changing methods, SI could be elevated.",
"SSVC technical_impact \u0027partial\u0027 is the least-assumptive choice; if cloud metadata endpoint access leads to full credential theft and infrastructure takeover, impact could be \u0027total\u0027.",
"The CAPEC-12 mapping is high-confidence based on the explicit SSRF description in the commit message and the code changes; no uncertainty is expected.",
"No specific affected or fixed version numbers are provided in the patch metadata; version boundaries are marked as unspecified.",
"The is_global check in Python\u0027s ipaddress module covers loopback, private, link-local, and other non-global ranges, but DNS rebinding between the check and the actual fetch is not addressed by this patch and could remain a residual risk."
],
"capecRationale": [
{
"capecId": "CAPEC-12",
"rationale": "The patch explicitly describes and remediates an SSRF: an attacker-supplied URL causes the MISP server to fetch internal resources (loopback, cloud metadata) and return the data. CAPEC-12 is the direct and most specific attack pattern for this scenario. No uncertainty is noted; the commit message and code changes align precisely with this pattern."
}
],
"commit": "7cb31e96a1f59bec9afbc087b27b26d9986c6841",
"confidence": "high",
"credits": [
{
"lang": "en",
"type": "remediation developer",
"value": "elhoim"
},
{
"lang": "en",
"type": "remediation developer",
"value": "Claude Opus 5 (1M context)"
}
],
"cvssRationale": "AV:N: The MISP API is network-accessible. AC:L: Crafting a URL attribute with an internal address is trivial. AT:N: No special adversary technique is required. PR:L: An authenticated MISP user with attribute-creation rights is needed. UI:N: No victim interaction beyond the attacker\u0027s own API call. VC/VI/VA:N: The MISP server\u0027s own confidentiality, integrity, and availability are not directly compromised. SC:H: The subsequent component (internal services, cloud metadata) suffers high confidentiality impact because its data is exfiltrated into the MISP event. SI:N: The module only performs GET-style fetches; no evidence of integrity impact on subsequent components. SA:N: No availability impact on subsequent components is indicated.",
"draft": false,
"fixSummary": "The fix introduces a URL validation gate (is_safe_url) that is invoked before any archiving request is made. The function restricts the scheme to http or https, resolves the hostname via DNS, and verifies that every resolved IP address is globally routable (i.e., not loopback, private, link-local, or otherwise non-global). If any check fails, the module returns an error and the fetch is never performed, preventing the server from being directed to internal or non-HTTP targets.",
"generatedAt": "2026-11-10T11:02:25.004287Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 7,
"confidence": "high",
"model": "qwen3.8:27b",
"score": 6
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "736974194693496cbcfe68e570862a9e221a43f5f6820180e20958b617ebb97c",
"patchSummary": "Adds imports for ipaddress and socket. Introduces a new is_safe_url(url) function that parses the URL, rejects non-http/https schemes, resolves the hostname with socket.getaddrinfo, and rejects the URL if any resolved address fails ipaddress.ip_address().is_global(). In the handler() function, a guard clause calls is_safe_url(tocache) before invoking __archiveLink and returns an error dict if the URL is not safe. Two new unit tests verify that http://127.0.0.1:6379/ and file:///etc/passwd are blocked and that the archive function is never called.",
"patchTruncated": false,
"patches": [
{
"commit": "7cb31e96a1f59bec9afbc087b27b26d9986c6841",
"date": "Fri, 16 Oct 2026 22:27:37 +0000",
"patchSha256": "736974194693496cbcfe68e570862a9e221a43f5f6820180e20958b617ebb97c",
"retrievedAt": "2026-11-10T11:01:44.851219Z",
"source": "https://github.com/MISP/misp-modules/commit/7cb31e96a1f59bec9afbc087b27b26d9986c6841.patch",
"sourceUrl": "https://github.com/MISP/misp-modules/commit/7cb31e96a1f59bec9afbc087b27b26d9986c6841.patch",
"subject": "fix: [sourcecache] validate URLs before archiving to block"
}
],
"source": "https://github.com/MISP/misp-modules/commit/7cb31e96a1f59bec9afbc087b27b26d9986c6841.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "partial"
}
],
"role": "Supplier",
"timestamp": "2026-11-10T11:02:25Z",
"version": "2.0.3"
},
"subject": "fix: [sourcecache] validate URLs before archiving to block",
"weaknessRationale": [
{
"cweId": "CWE-918",
"rationale": "The module accepts a user-supplied URL and issues a server-side HTTP request to it without validating the target, allowing an attacker to direct requests to internal or non-HTTP resources. This is a textbook SSRF."
},
{
"cweId": "CWE-20",
"rationale": "The root cause is the absence of any validation on the URL attribute value before it is used as a request target. The fix adds scheme, hostname, and IP-address validation, confirming the missing input validation was the defect."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"description": "Corrective change authored (7cb31e96a1f59bec9afbc087b27b26d9986c6841): fix: [sourcecache] validate URLs before archiving to block",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/MISP/misp-modules/commit/7cb31e96a1f59bec9afbc087b27b26d9986c6841.patch"
],
"timestamp": "2026-10-16T22:27:37Z",
"type": "fix-developed"
},
{
"description": "Patch retrieved from the supplied URL; retrieval time is used as the default availability time.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/MISP/misp-modules/commit/7cb31e96a1f59bec9afbc087b27b26d9986c6841.patch"
],
"timestamp": "2026-11-10T11:01:44.851219Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20273"
}
]
}
},
"cveMetadata": {
"cveId": "CVE-2026-125871",
"datePublished": "2026-11-10T12:48:56.140741Z",
"dateReserved": "2026-11-10T12:50:34.525Z",
"dateUpdated": "2026-11-10T15:01:59.744395Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20273"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2.0"
}
GCVE-1-2026-20156 (CVE-2026-124850)
Vulnerability from gna-1 – Published: 2026-11-09 14:32 – Updated: 2026-11-09 21:47- CWE-89 - Improper Neutralization of Special Elements used in an SQL Command ('SQL Injection')
| URL | Tags |
|---|---|
| https://github.com/flowintel/flowintel/commit/061… | patch |
qwen3.8:27b
2 events
advisory
bcp-05-x-01bcp-05-x-02bcp-05-x-03
Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.
| Model | Source | Identifier |
|---|---|---|
| qwen3.8:27b | ollama | qwen3.8:27b |
- Generator
-
patch2vuln.pyon 2026-11-09 14:12 - Model
qwen3.8:27b- Input
-
https://github.com/flowintel/flowintel/commit/061053689dc279a3c2c182f8dc15cd42dc24bd7d.patch
9ad956ded461… - Confidence
- medium
| Commit | Subject | Patch SHA-256 |
|---|---|---|
061053689dc2
|
fix: [security] Bind stored connector identifiers during | 9ad956ded461… |
Fix summary
The vulnerability is remediated by replacing direct string interpolation of stored database values into SQL queries with parameterized (bound) queries, ensuring that user-supplied or stored data is treated strictly as data rather than executable SQL. Additionally, the NULL handling for identifier fields is corrected to use proper SQL NULL instead of the string literal 'null', preventing both injection and unintended data corruption.
Patch summary
In migrations/versions/5cd03995339b_.py, four SQL statements (two SELECT and two UPDATE) are converted from Python f-string interpolation to SQLAlchemy parameterized queries using named bind parameters (:case_id, :instance_id, :ident, :id, :task_id). The NULL identifier handling is changed from assigning the string 'null' to assigning Python None, which SQLAlchemy translates to a proper SQL NULL value.
CVSS rationale
AV:L: The migration runs locally on the application server; exploitation requires local access to the database. AC:H: High complexity because the attacker must first write a crafted identifier into the database and then wait for a migration to execute. AT:P: Attack targeting is present; the attacker must specifically craft a malicious SQL payload in the identifier field. PR:L: Low privilege; the attacker needs write access to the database (e.g., an authenticated application user or DB access). UI:N: The migration runs automatically during upgrade; no user interaction is required at trigger time. VC:H, VI:H, VA:H: Successful SQL injection allows arbitrary read, write, and destructive operations on the database. SC:N, SI:N, SA:N: Impact is confined to the local database; no evidence of cross-component or cross-system impact.
Weakness rationale
- CWE-89 The patch directly replaces f-string interpolation of stored values into SQL with parameterized queries, which is the canonical fix for CWE-89. The commit message explicitly states that stored identifiers were 'interpolated into SQL, allowing expressions to execute,' confirming SQL injection.
Attack pattern rationale
- CAPEC-126 The vulnerability is a textbook SQL injection: attacker-controlled data (stored connector identifiers) is concatenated into SQL statements without sanitization or parameterization. CAPEC-126 is the most direct and specific match. The stored nature of the injection (data planted earlier, executed later during migration) is a variant but still falls squarely within this pattern.
Assumptions to verify
- The affected version range is inferred from the explicit context (affected_version '0', fixed_version '<=4.0.0'); the exact starting version is not confirmed by the patch itself.
- PR:L assumes the attacker can write to the database via an authenticated application role or direct DB access; the specific authentication mechanism is not stated in the patch.
- The CVSS impact ratings (VC:H, VI:H, VA:H) assume the database user has broad privileges; if the DB account is highly restricted, impact could be lower.
- CAPEC-126 is selected as the closest match; the stored/timed nature of the injection (data planted before migration runs) is a variant not explicitly covered by a more specific CAPEC.
- The patch does not indicate whether the migration has already been executed in deployed instances, which would affect real-world exploitability.
Model comparison
Selected qwen3.8:27b
by deterministic-consensus-v1
The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required.
| Model | Score | Agreement | Confidence | Assumptions |
|---|---|---|---|---|
qwen3.8:27b |
7 | 11 | medium | 5 |
- 2026-11-06 09:54 UTC Fix developed Corrective change authored (061053689dc279a3c2c182f8dc15cd42dc24bd7d): fix: [security] Bind stored connector identifiers during https://github.com/flowintel/flowintel/commit/061053689dc279a3c2c182f8dc15cd42dc24bd7d.patch
- 2026-11-09 14:10 UTC Fix available Patch retrieved from the supplied URL; retrieval time is used as the default availability time. https://github.com/flowintel/flowintel/commit/061053689dc279a3c2c182f8dc15cd42dc24bd7d.patch
{
"containers": {
"cna": {
"affected": [
{
"cpes": [
"cpe:2.3:a:flowintel:flowintel:*:*:*:*:*:*:*:*"
],
"modules": [
"migrations"
],
"product": "flowintel",
"programFiles": [
"migrations/versions/5cd03995339b_.py"
],
"repo": "https://github.com/flowintel/flowintel",
"vendor": "flowintel",
"versions": [
{
"lessThanOrEqual": "4.0.0",
"status": "affected",
"version": "0",
"versionType": "semver"
}
]
}
],
"cpeApplicability": [
{
"nodes": [
{
"cpeMatch": [
{
"criteria": "cpe:2.3:a:flowintel:flowintel:*:*:*:*:*:*:*:*",
"vulnerable": true
}
],
"negate": false,
"operator": "OR"
}
]
}
],
"credits": [
{
"lang": "en",
"type": "reporter",
"value": "elhoim"
},
{
"lang": "en",
"type": "remediation developer",
"value": "David Cruciani"
}
],
"descriptions": [
{
"lang": "en",
"value": "A SQL injection vulnerability exists in a database migration script in flowintel.\n\nThe migration process reads stored connector identifier values from the database and interpolates them directly into SQL statements without parameterization. An attacker who has previously written a malicious identifier value into the affected tables can cause arbitrary SQL execution when the migration is subsequently run during an application upgrade.\n\n- Requires prior write access to the database to plant a malicious identifier value\n- Triggered automatically during a migration or upgrade operation\n- Can result in unauthorized data disclosure, modification, or deletion\n\nAffected versions: flowintel prior to 4.0.0"
}
],
"impacts": [
{
"capecId": "CAPEC-126",
"descriptions": [
{
"lang": "en",
"value": "CAPEC-126 SQL Injection"
}
]
}
],
"metrics": [
{
"cvssV4_0": {
"Automatable": "NOT_DEFINED",
"Recovery": "NOT_DEFINED",
"Safety": "NOT_DEFINED",
"attackComplexity": "HIGH",
"attackRequirements": "PRESENT",
"attackVector": "LOCAL",
"baseScore": 7.3,
"baseSeverity": "HIGH",
"privilegesRequired": "LOW",
"providerUrgency": "NOT_DEFINED",
"subAvailabilityImpact": "NONE",
"subConfidentialityImpact": "NONE",
"subIntegrityImpact": "NONE",
"userInteraction": "NONE",
"valueDensity": "NOT_DEFINED",
"vectorString": "CVSS:4.0/AV:L/AC:H/AT:P/PR:L/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N",
"version": "4.0",
"vulnAvailabilityImpact": "HIGH",
"vulnConfidentialityImpact": "HIGH",
"vulnIntegrityImpact": "HIGH",
"vulnerabilityResponseEffort": "NOT_DEFINED"
},
"format": "CVSS",
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
},
{
"format": "SSVC",
"other": {
"content": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "total"
}
],
"role": "Supplier",
"timestamp": "2026-11-09T14:12:17Z",
"version": "2.0.3"
},
"type": "SSVC"
},
"scenarios": [
{
"lang": "en",
"value": "GENERAL"
}
]
}
],
"problemTypes": [
{
"descriptions": [
{
"cweId": "CWE-89",
"description": "CWE-89 Improper Neutralization of Special Elements used in an SQL Command (\u0027SQL Injection\u0027)",
"lang": "en",
"type": "CWE"
}
]
}
],
"providerMetadata": {
"orgId": "00000000-0000-4000-8000-000000000000"
},
"references": [
{
"name": "Security patch",
"tags": [
"patch"
],
"url": "https://github.com/flowintel/flowintel/commit/061053689dc279a3c2c182f8dc15cd42dc24bd7d"
}
],
"solutions": [
{
"lang": "en",
"value": "The vulnerability is remediated by replacing direct string interpolation of stored database values into SQL queries with parameterized (bound) queries, ensuring that user-supplied or stored data is treated strictly as data rather than executable SQL. Additionally, the NULL handling for identifier fields is corrected to use proper SQL NULL instead of the string literal \u0027null\u0027, preventing both injection and unintended data corruption."
}
],
"title": "SQL Injection in Database Migration via Stored Connector Identifiers in flowintel",
"x_gcve": [
{
"extensions": {
"bcp-05-x-01": {
"ai_annotations": [
{
"ai_level": "generated",
"description": "Draft vulnerability metadata was generated from a git-format patch using an Ollama-hosted language model. Human validation is required before publication.",
"gna_source": 1,
"models": [
{
"gna_source": 1,
"identifier": "qwen3.8:27b",
"name": "qwen3.8:27b",
"source": "ollama"
}
],
"review_status": "full",
"scope": "record",
"tags": [
"ai-computer-assisted:llm-generated",
"ai-computer-assisted:classification"
]
}
]
},
"bcp-05-x-02": {
"x_patch2vuln": {
"assumptions": [
"The affected version range is inferred from the explicit context (affected_version \u00270\u0027, fixed_version \u0027\u003c=4.0.0\u0027); the exact starting version is not confirmed by the patch itself.",
"PR:L assumes the attacker can write to the database via an authenticated application role or direct DB access; the specific authentication mechanism is not stated in the patch.",
"The CVSS impact ratings (VC:H, VI:H, VA:H) assume the database user has broad privileges; if the DB account is highly restricted, impact could be lower.",
"CAPEC-126 is selected as the closest match; the stored/timed nature of the injection (data planted before migration runs) is a variant not explicitly covered by a more specific CAPEC.",
"The patch does not indicate whether the migration has already been executed in deployed instances, which would affect real-world exploitability."
],
"capecRationale": [
{
"capecId": "CAPEC-126",
"rationale": "The vulnerability is a textbook SQL injection: attacker-controlled data (stored connector identifiers) is concatenated into SQL statements without sanitization or parameterization. CAPEC-126 is the most direct and specific match. The stored nature of the injection (data planted earlier, executed later during migration) is a variant but still falls squarely within this pattern."
}
],
"commit": "061053689dc279a3c2c182f8dc15cd42dc24bd7d",
"confidence": "medium",
"credits": [
{
"lang": "en",
"type": "reporter",
"value": "elhoim"
},
{
"lang": "en",
"type": "remediation developer",
"value": "David Cruciani"
}
],
"cvssRationale": "AV:L: The migration runs locally on the application server; exploitation requires local access to the database. AC:H: High complexity because the attacker must first write a crafted identifier into the database and then wait for a migration to execute. AT:P: Attack targeting is present; the attacker must specifically craft a malicious SQL payload in the identifier field. PR:L: Low privilege; the attacker needs write access to the database (e.g., an authenticated application user or DB access). UI:N: The migration runs automatically during upgrade; no user interaction is required at trigger time. VC:H, VI:H, VA:H: Successful SQL injection allows arbitrary read, write, and destructive operations on the database. SC:N, SI:N, SA:N: Impact is confined to the local database; no evidence of cross-component or cross-system impact.",
"fixSummary": "The vulnerability is remediated by replacing direct string interpolation of stored database values into SQL queries with parameterized (bound) queries, ensuring that user-supplied or stored data is treated strictly as data rather than executable SQL. Additionally, the NULL handling for identifier fields is corrected to use proper SQL NULL instead of the string literal \u0027null\u0027, preventing both injection and unintended data corruption.",
"generatedAt": "2026-11-09T14:12:17.714022Z",
"generator": "patch2vuln.py",
"model": "qwen3.8:27b",
"modelComparison": {
"potentialErrors": [],
"rankings": [
{
"agreementScore": 11,
"assumptionCount": 5,
"confidence": "medium",
"model": "qwen3.8:27b",
"score": 7
}
],
"selectedModel": "qwen3.8:27b",
"selectionMethod": "deterministic-consensus-v1",
"selectionNotice": "The selected result is closest to model consensus; this heuristic does not establish factual correctness and human review remains required."
},
"patchSha256": "9ad956ded461ca4c442070a86c76b0239017eeb48e36892684465b9b6fb4ccde",
"patchSummary": "In migrations/versions/5cd03995339b_.py, four SQL statements (two SELECT and two UPDATE) are converted from Python f-string interpolation to SQLAlchemy parameterized queries using named bind parameters (:case_id, :instance_id, :ident, :id, :task_id). The NULL identifier handling is changed from assigning the string \u0027null\u0027 to assigning Python None, which SQLAlchemy translates to a proper SQL NULL value.",
"patchTruncated": false,
"patches": [
{
"commit": "061053689dc279a3c2c182f8dc15cd42dc24bd7d",
"date": "Fri, 6 Nov 2026 10:54:39 +0100",
"patchSha256": "9ad956ded461ca4c442070a86c76b0239017eeb48e36892684465b9b6fb4ccde",
"retrievedAt": "2026-11-09T14:10:44.764240Z",
"source": "https://github.com/flowintel/flowintel/commit/061053689dc279a3c2c182f8dc15cd42dc24bd7d.patch",
"sourceUrl": "https://github.com/flowintel/flowintel/commit/061053689dc279a3c2c182f8dc15cd42dc24bd7d.patch",
"subject": "fix: [security] Bind stored connector identifiers during"
}
],
"source": "https://github.com/flowintel/flowintel/commit/061053689dc279a3c2c182f8dc15cd42dc24bd7d.patch",
"ssvc": {
"options": [
{
"Exploitation": "none"
},
{
"Automatable": "yes"
},
{
"Technical Impact": "total"
}
],
"role": "Supplier",
"timestamp": "2026-11-09T14:12:17Z",
"version": "2.0.3"
},
"subject": "fix: [security] Bind stored connector identifiers during",
"weaknessRationale": [
{
"cweId": "CWE-89",
"rationale": "The patch directly replaces f-string interpolation of stored values into SQL with parameterized queries, which is the canonical fix for CWE-89. The commit message explicitly states that stored identifiers were \u0027interpolated into SQL, allowing expressions to execute,\u0027 confirming SQL injection."
}
]
}
},
"bcp-05-x-03": {
"x_timeline": {
"events": [
{
"description": "Corrective change authored (061053689dc279a3c2c182f8dc15cd42dc24bd7d): fix: [security] Bind stored connector identifiers during",
"id": "evt-fix-developed-1",
"references": [
"https://github.com/flowintel/flowintel/commit/061053689dc279a3c2c182f8dc15cd42dc24bd7d.patch"
],
"timestamp": "2026-11-06T09:54:39Z",
"type": "fix-developed"
},
{
"description": "Patch retrieved from the supplied URL; retrieval time is used as the default availability time.",
"id": "evt-fix-available-1",
"references": [
"https://github.com/flowintel/flowintel/commit/061053689dc279a3c2c182f8dc15cd42dc24bd7d.patch"
],
"timestamp": "2026-11-09T14:10:44.764240Z",
"type": "fix-available"
}
]
}
}
},
"recordType": "advisory",
"vulnId": "gcve-1-2026-20156"
}
]
}
},
"cveMetadata": {
"assignerOrgId": "00000000-0000-4000-8000-000000000000",
"cveId": "CVE-2026-124850",
"datePublished": "2026-11-09T14:32:37.866692Z",
"dateReserved": "2026-11-09T14:38:45.169Z",
"dateUpdated": "2026-11-09T21:47:47.877339Z",
"state": "PUBLISHED",
"vulnId": "gcve-1-2026-20156"
},
"dataType": "CVE_RECORD",
"dataVersion": "5.2"
}