CWE-367
AllowedTime-of-check Time-of-use (TOCTOU) Race Condition
Abstraction: Base · Status: Incomplete
The product checks the state of a resource before using that resource, but the resource's state can change between the check and the use in a way that invalidates the results of the check.
1408 vulnerabilities reference this CWE, most recent first.
GHSA-Q5W7-QMX3-4VXH
Vulnerability from github – Published: 2026-01-13 18:31 – Updated: 2026-01-13 18:31Time-of-check time-of-use (toctou) race condition in Windows Ancillary Function Driver for WinSock allows an authorized attacker to elevate privileges locally.
{
"affected": [],
"aliases": [
"CVE-2026-20831"
],
"database_specific": {
"cwe_ids": [
"CWE-367"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-01-13T18:16:10Z",
"severity": "HIGH"
},
"details": "Time-of-check time-of-use (toctou) race condition in Windows Ancillary Function Driver for WinSock allows an authorized attacker to elevate privileges locally.",
"id": "GHSA-q5w7-qmx3-4vxh",
"modified": "2026-01-13T18:31:08Z",
"published": "2026-01-13T18:31:08Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-20831"
},
{
"type": "WEB",
"url": "https://msrc.microsoft.com/update-guide/vulnerability/CVE-2026-20831"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
GHSA-Q6M9-XJ2W-XMRC
Vulnerability from github – Published: 2026-04-22 18:31 – Updated: 2026-04-30 17:30The touch utility in uutils coreutils is vulnerable to a Time-of-Check to Time-of-Use (TOCTOU) race condition during file creation. When the utility identifies a missing path, it later attempts creation using File::create(), which internally uses O_TRUNC. An attacker can exploit this window to create a file or swap a symlink at the target path, causing touch to truncate an existing file and leading to permanent data loss.
{
"affected": [
{
"package": {
"ecosystem": "crates.io",
"name": "coreutils"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"last_affected": "0.8.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-35360"
],
"database_specific": {
"cwe_ids": [
"CWE-367"
],
"github_reviewed": true,
"github_reviewed_at": "2026-04-30T17:30:18Z",
"nvd_published_at": "2026-04-22T17:16:38Z",
"severity": "MODERATE"
},
"details": "The touch utility in uutils coreutils is vulnerable to a Time-of-Check to Time-of-Use (TOCTOU) race condition during file creation. When the utility identifies a missing path, it later attempts creation using File::create(), which internally uses O_TRUNC. An attacker can exploit this window to create a file or swap a symlink at the target path, causing touch to truncate an existing file and leading to permanent data loss.",
"id": "GHSA-q6m9-xj2w-xmrc",
"modified": "2026-04-30T17:30:18Z",
"published": "2026-04-22T18:31:45Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-35360"
},
{
"type": "WEB",
"url": "https://github.com/uutils/coreutils/issues/10019"
},
{
"type": "PACKAGE",
"url": "https://github.com/uutils/coreutils"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:N/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "uutils coreutils has a Time-of-check Time-of-use (TOCTOU) Race Condition"
}
GHSA-Q6QM-PXCX-GJ5H
Vulnerability from github – Published: 2022-01-22 00:00 – Updated: 2022-08-06 00:00A race condition was found in the Linux kernel's ebpf verifier between bpf_map_update_elem and bpf_map_freeze due to a missing lock in kernel/bpf/syscall.c. In this flaw, a local user with a special privilege (cap_sys_admin or cap_bpf) can modify the frozen mapped address space. This flaw affects kernel versions prior to 5.16 rc2.
{
"affected": [],
"aliases": [
"CVE-2021-4001"
],
"database_specific": {
"cwe_ids": [
"CWE-367"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2022-01-21T19:15:00Z",
"severity": "MODERATE"
},
"details": "A race condition was found in the Linux kernel\u0027s ebpf verifier between bpf_map_update_elem and bpf_map_freeze due to a missing lock in kernel/bpf/syscall.c. In this flaw, a local user with a special privilege (cap_sys_admin or cap_bpf) can modify the frozen mapped address space. This flaw affects kernel versions prior to 5.16 rc2.",
"id": "GHSA-q6qm-pxcx-gj5h",
"modified": "2022-08-06T00:00:53Z",
"published": "2022-01-22T00:00:22Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2021-4001"
},
{
"type": "WEB",
"url": "https://bugzilla.redhat.com/show_bug.cgi?id=2025645"
},
{
"type": "WEB",
"url": "https://git.kernel.org/pub/scm/linux/kernel/git/bpf/bpf.git/commit/?id=353050be4c19e102178ccc05988101887c25ae53"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:U/C:N/I:H/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-Q86M-697P-H7FH
Vulnerability from github – Published: 2026-03-19 03:30 – Updated: 2026-03-19 16:24Duplicate Advisory
This advisory has been withdrawn because it is a duplicate of GHSA-q399-23r3-hfx4. This link is maintained to preserve external references.
Original Description
OpenClaw versions prior to 2026.3.1 fail to pin executable identity for non-path-like argv[0] tokens in system.run approvals, allowing post-approval executable rebind attacks. Attackers can modify PATH resolution after approval to execute a different binary than the operator approved, enabling arbitrary command execution.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c 2026.3.1"
},
"package": {
"ecosystem": "npm",
"name": "openclaw"
},
"ranges": [
{
"events": [
{
"introduced": "0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-367"
],
"github_reviewed": true,
"github_reviewed_at": "2026-03-19T16:24:36Z",
"nvd_published_at": "2026-03-19T02:16:05Z",
"severity": "MODERATE"
},
"details": "## Duplicate Advisory\n\nThis advisory has been withdrawn because it is a duplicate of GHSA-q399-23r3-hfx4. This link is maintained to preserve external references.\n\n## Original Description\nOpenClaw versions prior to 2026.3.1 fail to pin executable identity for non-path-like argv[0] tokens in system.run approvals, allowing post-approval executable rebind attacks. Attackers can modify PATH resolution after approval to execute a different binary than the operator approved, enabling arbitrary command execution.",
"id": "GHSA-q86m-697p-h7fh",
"modified": "2026-03-19T16:24:36Z",
"published": "2026-03-19T03:30:57Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/openclaw/openclaw/security/advisories/GHSA-q399-23r3-hfx4"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-31997"
},
{
"type": "WEB",
"url": "https://www.vulncheck.com/advisories/openclaw-executable-rebind-via-unbound-path-token-in-system-run-approvals"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:H/PR:L/UI:R/S:U/C:H/I:H/A:N",
"type": "CVSS_V3"
},
{
"score": "CVSS:4.0/AV:L/AC:H/AT:N/PR:L/UI:A/VC:N/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X",
"type": "CVSS_V4"
}
],
"summary": "Duplicate Advisory: OpenClaw: system.run approvals did not bind PATH-token executable identity, enabling post-approval executable rebind",
"withdrawn": "2026-03-19T16:24:36Z"
}
GHSA-Q945-MJ3H-45F2
Vulnerability from github – Published: 2023-04-25 00:30 – Updated: 2024-01-19 18:30The specific flaw exists within the DPT I2O Controller driver. The issue results from the lack of proper locking when performing operations on an object. An attacker can leverage this in conjunction with other vulnerabilities to escalate privileges and execute arbitrary code in the context of the kernel.
{
"affected": [],
"aliases": [
"CVE-2023-2007"
],
"database_specific": {
"cwe_ids": [
"CWE-367",
"CWE-667"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2023-04-24T23:15:18Z",
"severity": "HIGH"
},
"details": "The specific flaw exists within the DPT I2O Controller driver. The issue results from the lack of proper locking when performing operations on an object. An attacker can leverage this in conjunction with other vulnerabilities to escalate privileges and execute arbitrary code in the context of the kernel.",
"id": "GHSA-q945-mj3h-45f2",
"modified": "2024-01-19T18:30:22Z",
"published": "2023-04-25T00:30:41Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2023-2007"
},
{
"type": "WEB",
"url": "https://github.com/torvalds/linux/commit/b04e75a4a8a81887386a0d2dbf605a48e779d2a0"
},
{
"type": "WEB",
"url": "https://lists.debian.org/debian-lts-announce/2023/07/msg00030.html"
},
{
"type": "WEB",
"url": "https://lists.debian.org/debian-lts-announce/2023/10/msg00027.html"
},
{
"type": "WEB",
"url": "https://security.netapp.com/advisory/ntap-20240119-0011"
},
{
"type": "WEB",
"url": "https://www.debian.org/security/2023/dsa-5480"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
GHSA-QCC4-P59M-P54M
Vulnerability from github – Published: 2026-03-12 14:21 – Updated: 2026-03-12 14:21Summary
A sandbox boundary-validation gap in symlink alias handling allowed certain workspace-only write paths to be treated as in-boundary even when they could resolve outside the workspace/sandbox root.
Affected Packages / Versions
- Package: npm
openclaw - Affected versions:
<= 2026.2.25 - Latest published npm version included in affected range:
2026.2.25(checked on February 26, 2026) - Patched version (pre-set for release):
2026.2.26
Technical Details
In affected versions, dangling symlink hops could be accepted during boundary checks under missing-target conditions. For workspace-only write flows (including apply_patch), this could allow writes to resolve outside the configured workspace/sandbox boundary.
The fix resolves symlink targets through existing ancestors and fails closed when canonical resolution escapes the configured boundary.
Impact
- Boundary-confined write operations could be redirected outside the configured workspace/sandbox root.
- Primary impact is integrity of host-side files reachable from that path resolution.
Fix Commit(s)
4fd29a35bb85a1898ebff518364c467058b50e14
Release Process Note
patched_versions is pre-set to the planned next release (2026.2.26) so once npm 2026.2.26 is published, the advisory can be published without further field edits.
Thanks @tdjackey for reporting.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 2026.2.25"
},
"package": {
"ecosystem": "npm",
"name": "openclaw"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "2026.2.26"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-367",
"CWE-59"
],
"github_reviewed": true,
"github_reviewed_at": "2026-03-12T14:21:54Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "### Summary\nA sandbox boundary-validation gap in symlink alias handling allowed certain workspace-only write paths to be treated as in-boundary even when they could resolve outside the workspace/sandbox root.\n\n### Affected Packages / Versions\n- Package: npm `openclaw`\n- Affected versions: `\u003c= 2026.2.25`\n- Latest published npm version included in affected range: `2026.2.25` (checked on February 26, 2026)\n- Patched version (pre-set for release): `2026.2.26`\n\n### Technical Details\nIn affected versions, dangling symlink hops could be accepted during boundary checks under missing-target conditions. For workspace-only write flows (including `apply_patch`), this could allow writes to resolve outside the configured workspace/sandbox boundary.\n\nThe fix resolves symlink targets through existing ancestors and fails closed when canonical resolution escapes the configured boundary.\n\n### Impact\n- Boundary-confined write operations could be redirected outside the configured workspace/sandbox root.\n- Primary impact is integrity of host-side files reachable from that path resolution.\n\n### Fix Commit(s)\n- `4fd29a35bb85a1898ebff518364c467058b50e14`\n\n### Release Process Note\n`patched_versions` is pre-set to the planned next release (`2026.2.26`) so once npm `2026.2.26` is published, the advisory can be published without further field edits.\n\nThanks @tdjackey for reporting.",
"id": "GHSA-qcc4-p59m-p54m",
"modified": "2026-03-12T14:21:54Z",
"published": "2026-03-12T14:21:54Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/openclaw/openclaw/security/advisories/GHSA-qcc4-p59m-p54m"
},
{
"type": "WEB",
"url": "https://github.com/openclaw/openclaw/commit/4fd29a35bb85a1898ebff518364c467058b50e14"
},
{
"type": "PACKAGE",
"url": "https://github.com/openclaw/openclaw"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:H/PR:L/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "OpenClaw: Sandbox dangling-symlink alias handling could bypass workspace-only write boundary"
}
GHSA-QCHC-32XG-C7W6
Vulnerability from github – Published: 2026-09-14 15:32 – Updated: 2026-09-14 15:32Description
The setuid-root worker-launcher binary adjusts ownership and permissions of worker directories by walking
the tree with FTS and calling lchown and chmod on each entry's full pathname while running with an
effective uid of 0. Both syscalls re-resolve the path at the time of the call, after FTS has classified the
entry, and the trees being walked are owned and writable by the untrusted topology user.
A tenant running code on a supervisor node could therefore replace an intermediate directory component with
a symbolic link between classification and the privileged operation, redirecting the root-owned lchown or
chmod at an arbitrary file on the host. The operation is repeatable at will, since crashing a worker
forces a relaunch and blob updates re-run the walk, so a failed attempt costs the attacker nothing.
This crosses the boundary that supervisor.run.worker.as.user and container isolation are intended to
enforce. It is the same defect class as the Hadoop container-executor issues from which this code derives.
Mitigation
Upgrade to 3.1.0, where the privileged walk operates on file descriptors it has already stat'd rather than on pathnames re-resolved at call time.
Users who cannot upgrade immediately should not run untrusted topology code on supervisors configured with
supervisor.run.worker.as.user, since the launcher is the boundary being crossed. Note that the launcher
must be rebuilt and reinstalled after upgrading; replacing the Java artifacts alone is not sufficient.
Credit
The ASF -- found using Claude agents to study the security of open-source projects, validated and reported by Apache Storm.
{
"affected": [],
"aliases": [
"CVE-2026-82429"
],
"database_specific": {
"cwe_ids": [
"CWE-367"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-09-14T15:17:09Z",
"severity": null
},
"details": "Description\n\nThe setuid-root `worker-launcher` binary adjusts ownership and permissions of worker directories by walking\nthe tree with FTS and calling `lchown` and `chmod` on each entry\u0027s full pathname while running with an\neffective uid of 0. Both syscalls re-resolve the path at the time of the call, after FTS has classified the\nentry, and the trees being walked are owned and writable by the untrusted topology user.\n\nA tenant running code on a supervisor node could therefore replace an intermediate directory component with\na symbolic link between classification and the privileged operation, redirecting the root-owned `lchown` or\n`chmod` at an arbitrary file on the host. The operation is repeatable at will, since crashing a worker\nforces a relaunch and blob updates re-run the walk, so a failed attempt costs the attacker nothing.\n\nThis crosses the boundary that `supervisor.run.worker.as.user` and container isolation are intended to\nenforce. It is the same defect class as the Hadoop container-executor issues from which this code derives.\n\nMitigation\n\nUpgrade to 3.1.0, where the privileged walk operates on file descriptors it has already stat\u0027d rather than\non pathnames re-resolved at call time.\n\nUsers who cannot upgrade immediately should not run untrusted topology code on supervisors configured with\n`supervisor.run.worker.as.user`, since the launcher is the boundary being crossed. Note that the launcher\nmust be rebuilt and reinstalled after upgrading; replacing the Java artifacts alone is not sufficient.\n\nCredit\n\nThe ASF -- found using Claude agents to study the security of open-source projects, validated and reported by Apache Storm.",
"id": "GHSA-qchc-32xg-c7w6",
"modified": "2026-09-14T15:32:49Z",
"published": "2026-09-14T15:32:49Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-82429"
},
{
"type": "WEB",
"url": "https://lists.apache.org/thread/9vjqrt28g0mxg304vw6pw46cx72v4rvq"
}
],
"schema_version": "1.4.0",
"severity": []
}
GHSA-QF6V-76FQ-P8R9
Vulnerability from github – Published: 2023-08-07 06:30 – Updated: 2024-04-04 06:34In thermal, there is a possible use after free due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07648734; Issue ID: ALPS07648734.
{
"affected": [],
"aliases": [
"CVE-2023-20787"
],
"database_specific": {
"cwe_ids": [
"CWE-367"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2023-08-07T04:15:13Z",
"severity": "MODERATE"
},
"details": "In thermal, there is a possible use after free due to a race condition. This could lead to local escalation of privilege with System execution privileges needed. User interaction is not needed for exploitation. Patch ID: ALPS07648734; Issue ID: ALPS07648734.",
"id": "GHSA-qf6v-76fq-p8r9",
"modified": "2024-04-04T06:34:51Z",
"published": "2023-08-07T06:30:26Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2023-20787"
},
{
"type": "WEB",
"url": "https://corp.mediatek.com/product-security-bulletin/August-2023"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:H/PR:H/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
GHSA-QFJH-MFRQ-8J64
Vulnerability from github – Published: 2026-03-05 15:30 – Updated: 2026-04-01 15:30Avira Internet Security contains a time-of-check time-of-use (TOCTOU) vulnerability in the Optimizer component. A privileged service running as SYSTEM identifies directories for cleanup during a scan phase and subsequently deletes them during a separate cleanup phase without revalidating the target path. A local attacker can replace a previously scanned directory with a junction or reparse point before deletion occurs, causing the privileged process to delete an unintended system location. This may result in deletion of protected files or directories and can lead to local privilege escalation, denial of service, or system integrity compromise depending on the affected target.
{
"affected": [],
"aliases": [
"CVE-2026-27750"
],
"database_specific": {
"cwe_ids": [
"CWE-367"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-03-05T15:16:12Z",
"severity": "HIGH"
},
"details": "Avira Internet Security contains a time-of-check time-of-use (TOCTOU) vulnerability in the Optimizer component. A privileged service running as SYSTEM identifies directories for cleanup during a scan phase and subsequently deletes them during a separate cleanup phase without revalidating the target path. A local attacker can replace a previously scanned directory with a junction or reparse point before deletion occurs, causing the privileged process to delete an unintended system location. This may result in deletion of protected files or directories and can lead to local privilege escalation, denial of service, or system integrity compromise depending on the affected target.",
"id": "GHSA-qfjh-mfrq-8j64",
"modified": "2026-04-01T15:30:56Z",
"published": "2026-03-05T15:30:36Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-27750"
},
{
"type": "WEB",
"url": "https://blog.quarkslab.com/avira-deserialize-delete-and-escalate-the-proper-way-to-use-an-av.html"
},
{
"type": "WEB",
"url": "https://support.avira.com/hc/en-us/articles/360010656158-Current-Avira-versions"
},
{
"type": "WEB",
"url": "https://www.avira.com/en/internet-security"
},
{
"type": "WEB",
"url": "https://www.gendigital.com/us/en/contact-us/security-advisories"
},
{
"type": "WEB",
"url": "https://www.vulncheck.com/advisories/avira-internet-security-optimizer-toctou"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:L/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
},
{
"score": "CVSS:4.0/AV:L/AC:L/AT:N/PR:L/UI:N/VC:H/VI:H/VA:H/SC:N/SI:N/SA:N/E:X/CR:X/IR:X/AR:X/MAV:X/MAC:X/MAT:X/MPR:X/MUI:X/MVC:X/MVI:X/MVA:X/MSC:X/MSI:X/MSA:X/S:X/AU:X/R:X/V:X/RE:X/U:X",
"type": "CVSS_V4"
}
]
}
GHSA-QG25-6GC4-48MG
Vulnerability from github – Published: 2026-10-08 16:36 – Updated: 2026-10-08 16:36Summary
PraisonAI's web_crawl agent tool performs a server-side HTTP fetch of an agent/attacker-influenced URL. SSRF is meant to be prevented by _is_safe_crawl_url(), which resolves the hostname and rejects private/loopback/link-local IPs at validation time. The validated value is the URL string (not a pinned IP); the fetch backend then re-resolves the hostname at connection time. Because validation and connection perform two independent DNS resolutions, a DNS-rebinding domain that returns a public IP during validation and an internal IP during the fetch fully bypasses the guard, and the internal HTTP response body is returned to the caller.
This is SSRF with internal response disclosure (read-back) — not blind SSRF. Runtime-confirmed against PraisonAI 4.6.63; the crawl response returned the controlled internal markers PRAISONAI_INTERNAL_SECRET_CANARY_7f3a91 / FAKE_INTERNAL_TOKEN_DO_NOT_USE_7f3a91. Severity High. Reachable by any actor who can influence the URL an agent crawls (e.g. a chat/bot/agent surface).
Details
Affected component
- Package:
praisonaiagents(PraisonAI), version 4.6.63. - File:
src/praisonai-agents/praisonaiagents/tools/web_crawl_tools.py; toolweb_crawl/crawl_web(part of the default bot tool set).
Vulnerable code / root cause
Code point 1 — check-time-only DNS validation, no IP pinning
Path:
src/praisonai-agents/praisonaiagents/tools/web_crawl_tools.py
Function:
_is_safe_crawl_url
Snippet:
for info in socket.getaddrinfo(hostname, None): # resolve at CHECK time
ip = ipaddress.ip_address(info[4][0])
if (ip.is_loopback or ip.is_private or ip.is_link_local
or ip.is_multicast or ip.is_unspecified):
return False
return True
Issue: the guard validates the hostname by resolving it once at check time. It does not pin the resolved IP and does not return/forward that IP to the HTTP client. Any later resolution can differ.
Code point 2 — guard runs, then the URL string is handed to the backend
Function:
web_crawl
Snippet:
for u in raw_url_list:
if _is_safe_crawl_url(u): # validate the URL string
url_list.append(u)
...
results = _crawl_with_httpx(url_list) # or _crawl_with_crawl4ai(url_list)
Issue: attacker-controlled input (urls) is validated as a string; the backend then fetches that string and re-resolves DNS independently of the guard. There is no shared, pinned IP between check and fetch.
Code point 3 — _crawl_with_httpx backend re-resolves (redirect re-validation does not stop rebinding)
Function:
_crawl_with_httpx
Snippet:
with httpx.Client(follow_redirects=False, timeout=30.0) as client:
for _ in range(max_redirects + 1):
if not _is_safe_crawl_url(current): # re-resolves hostname (CHECK)
raise ValueError("Redirect target failed SSRF validation")
response = client.get(current) # resolves AGAIN at CONNECT
Issue: even with per-hop redirect re-validation, _is_safe_crawl_url(current) and client.get(current) are two separate DNS resolutions of the same hostname. A rebinding domain answers public to the check and internal to the connect → TOCTOU bypass. No IP pinning.
Code point 4 — urllib fallback (same function), no per-hop guard
Snippet:
import urllib.request
with urllib.request.urlopen(url, timeout=30) as response: # re-resolves + auto-follows redirects
content = response.read().decode('utf-8', errors='ignore')
Issue: when httpx is not installed, this fallback inside _crawl_with_httpx fetches the URL and auto-follows redirects with no per-hop/per-connect validation. (Results from this function are labelled "provider": "httpx" regardless of which path runs.)
Code point 5 — crawl4ai/Chromium backend (confirmed addendum)
The crawl4ai backend (_crawl_with_crawl4ai → crawler.arun(url=url), headless Chromium) is also runtime-confirmed affected (browser re-resolves DNS / follows redirects with no per-connect guard). To keep this report focused on the web_crawl SSRF guard, the backend-specific evidence is in SSRF-04_Crawl4AI_SSRF_Backend_Addendum.md.
Attack flow
- Attacker controls a hostname (e.g.
rebind.lab) whose authoritative DNS rebinds. - Lookup #1 (the guard) → a public IP →
_is_safe_crawl_url()returns true. - The backend re-resolves → the attacker's DNS now answers an internal/private IP (cloud metadata, loopback, internal service).
- The backend connects to the internal service and returns its body to the caller → internal data disclosure.
Why existing protection is bypassed
- The guard validates the hostname, not a pinned IP; check and connect resolve independently → DNS rebinding (TOCTOU) defeats it on every backend.
- Redirect re-validation (httpx path) re-checks the hostname but still re-resolves at connect, so it does not stop rebinding; the urllib fallback and crawl4ai backends have no per-hop guard at all.
Security boundary
The server-side fetch reaches internal/loopback/metadata services not exposed to the attacker and returns their content (CVSS Scope: Changed). Reachable wherever an agent can be induced to crawl an attacker-supplied URL (PR:L). An unauthenticated single-request path to web_crawl read-back was not found in 4.6.63 (so PR:N / Critical is not claimed).
Proof of Concept
Environment
Real PraisonAI 4.6.63 in a local Docker runtime; a controlled internal canary service (Docker-internal only, not published) returns synthetic markers; a controlled DNS responder implements rebinding for rebind.lab. No public host / real metadata / real secret. Runnable assets: PraisonAI-Runtime-Repro\runtime-files\.
Steps to reproduce
- Burp Repeater tab
PRAI-05-01-DNS-Rebind-Trigger→127.0.0.1:18080:
POST /tool/web_crawl HTTP/1.1
Host: 127.0.0.1:18080
Content-Type: application/json
{"url":"http://rebind.lab:8081/secret"}
- Send (
PRAI-05-02-DNS-Rebind-Secret-Readbackcaptures the response). If a send returns the "blocked" error, the rebinding DNS auto-resets (~3s) — resend. - Redirect variant:
PRAI-05-03-Redirect-Trigger/PRAI-05-04-Redirect-Secret-Readbacksend{"url":"http://redirector:8082/redirect-to-internal"}.
Expected result
A safe SSRF guard refuses destinations that resolve to internal/private IPs regardless of DNS timing or redirects, and does not return internal content.
Actual result
HTTP 200 with the internal body in the crawl result. Primary evidence is the provider: "httpx" backend returning the internal canary via DNS rebinding:
{"input_url":"http://rebind.lab:8081/secret",
"result":{"content":"{ ... \"secret\": \"PRAISONAI_INTERNAL_SECRET_CANARY_7f3a91\", \"token\": \"FAKE_INTERNAL_TOKEN_DO_NOT_USE_7f3a91\" ... }","provider":"httpx"}}
The redirect variant returns the same internal markers via a redirect chain (provider: "httpx").
Screenshots
DNS rebinding read-back
The attacker-controlled rebind.lab URL is accepted by web_crawl, and the PraisonAI response contains the internal canary response body.
DNS rebinding runtime evidence
The runtime log shows rebind.lab first resolving to an allowed/public IP during validation (guard-pass), then resolving to an internal Docker IP during the actual fetch (fetch-hit). The internal canary receives GET /secret from the PraisonAI container.
Redirect-based SSRF read-back
The attacker-controlled redirector URL is accepted by web_crawl. PraisonAI follows the redirect and returns the internal canary response body containing PRAISONAI_INTERNAL_SECRET_CANARY_7f3a91.
Redirect chain runtime evidence
The controlled redirector returns 302 -> http://internal-canary:8081/secret, and the internal canary receives GET /secret, confirming that the server-side client followed the redirect into the internal network.
Reproduction assets
The attached archive contains the local Docker runtime used to reproduce the issue with controlled canary services only. It does not contain real secrets, real cloud metadata access, or third-party API keys.
Impact
SSRF against internal/loopback/cloud-metadata endpoints with disclosure of internal HTTP responses (read-back) to the attacker. Bypasses the project's SSRF protection on every fetch backend.
Suggested remediation
- Resolve the host once, reject all returned records that are private/loopback/link-local/ULA/CGNAT/metadata, then connect to that exact validated IP (pin it; send the original
Host). Do not let the HTTP client / browser re-resolve. - Apply the same validation + IP pinning to every backend (httpx, urllib fallback, crawl4ai) and every redirect hop.
- Disable automatic redirect following (or cap + re-validate each hop with pinning).
- Treat IPv4-mapped IPv6, decimal/octal/hex IPs, and CGNAT/non-global ranges as unsafe.
{
"affected": [
{
"database_specific": {
"last_known_affected_version_range": "\u003c= 1.6.77"
},
"package": {
"ecosystem": "PyPI",
"name": "praisonaiagents"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "1.6.78"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-61430"
],
"database_specific": {
"cwe_ids": [
"CWE-200",
"CWE-367",
"CWE-918"
],
"github_reviewed": true,
"github_reviewed_at": "2026-10-08T16:36:50Z",
"nvd_published_at": "2026-07-15T17:16:52Z",
"severity": "HIGH"
},
"details": "## Summary\n\nPraisonAI\u0027s `web_crawl` agent tool performs a server-side HTTP fetch of an agent/attacker-influenced URL. SSRF is meant to be prevented by `_is_safe_crawl_url()`, which resolves the hostname and rejects private/loopback/link-local IPs **at validation time**. The validated value is the URL *string* (not a pinned IP); the fetch backend then **re-resolves the hostname at connection time**. Because validation and connection perform two independent DNS resolutions, a **DNS-rebinding** domain that returns a public IP during validation and an internal IP during the fetch fully bypasses the guard, and the internal HTTP response body is returned to the caller.\n\nThis is **SSRF with internal response disclosure (read-back)** \u2014 not blind SSRF. Runtime-confirmed against PraisonAI 4.6.63; the crawl response returned the controlled internal markers `PRAISONAI_INTERNAL_SECRET_CANARY_7f3a91` / `FAKE_INTERNAL_TOKEN_DO_NOT_USE_7f3a91`. Severity High. Reachable by any actor who can influence the URL an agent crawls (e.g. a chat/bot/agent surface).\n\n## Details\n\n### Affected component\n- Package: `praisonaiagents` (PraisonAI), version **4.6.63**.\n- File: `src/praisonai-agents/praisonaiagents/tools/web_crawl_tools.py`; tool `web_crawl` / `crawl_web` (part of the default bot tool set).\n\n### Vulnerable code / root cause\n\n**Code point 1 \u2014 check-time-only DNS validation, no IP pinning**\n\nPath:\n`src/praisonai-agents/praisonaiagents/tools/web_crawl_tools.py`\n\nFunction:\n`_is_safe_crawl_url`\n\nSnippet:\n```python\nfor info in socket.getaddrinfo(hostname, None): # resolve at CHECK time\n ip = ipaddress.ip_address(info[4][0])\n if (ip.is_loopback or ip.is_private or ip.is_link_local\n or ip.is_multicast or ip.is_unspecified):\n return False\nreturn True\n```\nIssue: the guard validates the hostname by resolving it **once at check time**. It does not pin the resolved IP and does not return/forward that IP to the HTTP client. Any later resolution can differ.\n\n**Code point 2 \u2014 guard runs, then the URL *string* is handed to the backend**\n\nFunction:\n`web_crawl`\n\nSnippet:\n```python\nfor u in raw_url_list:\n if _is_safe_crawl_url(u): # validate the URL string\n url_list.append(u)\n...\nresults = _crawl_with_httpx(url_list) # or _crawl_with_crawl4ai(url_list)\n```\nIssue: attacker-controlled input (`urls`) is validated as a string; the backend then fetches that string and **re-resolves DNS independently** of the guard. There is no shared, pinned IP between check and fetch.\n\n**Code point 3 \u2014 `_crawl_with_httpx` backend re-resolves (redirect re-validation does not stop rebinding)**\n\nFunction:\n`_crawl_with_httpx`\n\nSnippet:\n```python\nwith httpx.Client(follow_redirects=False, timeout=30.0) as client:\n for _ in range(max_redirects + 1):\n if not _is_safe_crawl_url(current): # re-resolves hostname (CHECK)\n raise ValueError(\"Redirect target failed SSRF validation\")\n response = client.get(current) # resolves AGAIN at CONNECT\n```\nIssue: even with per-hop redirect re-validation, `_is_safe_crawl_url(current)` and `client.get(current)` are **two separate DNS resolutions** of the same hostname. A rebinding domain answers public to the check and internal to the connect \u2192 TOCTOU bypass. No IP pinning.\n\n**Code point 4 \u2014 urllib fallback (same function), no per-hop guard**\n\nSnippet:\n```python\nimport urllib.request\nwith urllib.request.urlopen(url, timeout=30) as response: # re-resolves + auto-follows redirects\n content = response.read().decode(\u0027utf-8\u0027, errors=\u0027ignore\u0027)\n```\nIssue: when `httpx` is not installed, this fallback inside `_crawl_with_httpx` fetches the URL and auto-follows redirects with no per-hop/per-connect validation. (Results from this function are labelled `\"provider\": \"httpx\"` regardless of which path runs.)\n\n**Code point 5 \u2014 crawl4ai/Chromium backend (confirmed addendum)**\n\nThe crawl4ai backend (`_crawl_with_crawl4ai` \u2192 `crawler.arun(url=url)`, headless Chromium) is also runtime-confirmed affected (browser re-resolves DNS / follows redirects with no per-connect guard). To keep this report focused on the `web_crawl` SSRF guard, the backend-specific evidence is in `SSRF-04_Crawl4AI_SSRF_Backend_Addendum.md`.\n\n### Attack flow\n1. Attacker controls a hostname (e.g. `rebind.lab`) whose authoritative DNS rebinds.\n2. Lookup #1 (the guard) \u2192 a public IP \u2192 `_is_safe_crawl_url()` returns true.\n3. The backend re-resolves \u2192 the attacker\u0027s DNS now answers an internal/private IP (cloud metadata, loopback, internal service).\n4. The backend connects to the internal service and returns its body to the caller \u2192 internal data disclosure.\n\n### Why existing protection is bypassed\n- The guard validates the hostname, not a pinned IP; check and connect resolve independently \u2192 DNS rebinding (TOCTOU) defeats it on every backend.\n- Redirect re-validation (httpx path) re-checks the *hostname* but still re-resolves at connect, so it does not stop rebinding; the urllib fallback and crawl4ai backends have no per-hop guard at all.\n\n### Security boundary\nThe server-side fetch reaches internal/loopback/metadata services not exposed to the attacker and returns their content (CVSS Scope: Changed). Reachable wherever an agent can be induced to crawl an attacker-supplied URL (PR:L). An unauthenticated single-request path to `web_crawl` read-back was not found in 4.6.63 (so PR:N / Critical is not claimed).\n\n## Proof of Concept\n\n### Environment\nReal PraisonAI 4.6.63 in a local Docker runtime; a controlled internal canary service (Docker-internal only, not published) returns synthetic markers; a controlled DNS responder implements rebinding for `rebind.lab`. No public host / real metadata / real secret. Runnable assets: `PraisonAI-Runtime-Repro\\runtime-files\\`.\n\n### Steps to reproduce\n1. Burp Repeater tab `PRAI-05-01-DNS-Rebind-Trigger` \u2192 `127.0.0.1:18080`:\n```http\nPOST /tool/web_crawl HTTP/1.1\nHost: 127.0.0.1:18080\nContent-Type: application/json\n\n{\"url\":\"http://rebind.lab:8081/secret\"}\n```\n2. Send (`PRAI-05-02-DNS-Rebind-Secret-Readback` captures the response). If a send returns the \"blocked\" error, the rebinding DNS auto-resets (~3s) \u2014 resend.\n3. Redirect variant: `PRAI-05-03-Redirect-Trigger` / `PRAI-05-04-Redirect-Secret-Readback` send `{\"url\":\"http://redirector:8082/redirect-to-internal\"}`.\n\n### Expected result\nA safe SSRF guard refuses destinations that resolve to internal/private IPs regardless of DNS timing or redirects, and does not return internal content.\n\n### Actual result\nHTTP 200 with the internal body in the crawl result. Primary evidence is the `provider: \"httpx\"` backend returning the internal canary via DNS rebinding:\n```json\n{\"input_url\":\"http://rebind.lab:8081/secret\",\n \"result\":{\"content\":\"{ ... \\\"secret\\\": \\\"PRAISONAI_INTERNAL_SECRET_CANARY_7f3a91\\\", \\\"token\\\": \\\"FAKE_INTERNAL_TOKEN_DO_NOT_USE_7f3a91\\\" ... }\",\"provider\":\"httpx\"}}\n```\nThe redirect variant returns the same internal markers via a redirect chain (`provider: \"httpx\"`).\n\n### Screenshots\n\n**DNS rebinding read-back**\n\nThe attacker-controlled `rebind.lab` URL is accepted by `web_crawl`, and the PraisonAI response contains the internal canary response body.\n\n\u003cimg width=\"1543\" height=\"785\" alt=\"01-DNS-Rebind-Burp-Readback\" src=\"https://github.com/user-attachments/assets/e86e95dd-3d3a-4bef-b1c6-cb897234a212\" /\u003e\n\n**DNS rebinding runtime evidence**\n\nThe runtime log shows `rebind.lab` first resolving to an allowed/public IP during validation (`guard-pass`), then resolving to an internal Docker IP during the actual fetch (`fetch-hit`). The internal canary receives `GET /secret` from the PraisonAI container.\n\n\u003cimg width=\"1654\" height=\"828\" alt=\"02-DNS-Rebind-DNS-Log-And-Internal-Hit\" src=\"https://github.com/user-attachments/assets/f1d28046-de34-48f0-9bee-bbe26e598d5f\" /\u003e\n\n**Redirect-based SSRF read-back**\n\nThe attacker-controlled redirector URL is accepted by `web_crawl`. PraisonAI follows the redirect and returns the internal canary response body containing `PRAISONAI_INTERNAL_SECRET_CANARY_7f3a91`.\n\n\u003cimg width=\"1540\" height=\"772\" alt=\"03-Redirect-Burp-Readback\" src=\"https://github.com/user-attachments/assets/79eec159-4b9f-44be-a9eb-b15aba35ead8\" /\u003e\n\n**Redirect chain runtime evidence**\n\nThe controlled redirector returns `302 -\u003e http://internal-canary:8081/secret`, and the internal canary receives `GET /secret`, confirming that the server-side client followed the redirect into the internal network.\n\n\u003cimg width=\"1637\" height=\"894\" alt=\"04-Redirect-Internal-Hit-Log\" src=\"https://github.com/user-attachments/assets/1c503e30-9d01-4d32-8658-497f755a969e\" /\u003e\n\n### Reproduction assets\n\nThe attached archive contains the local Docker runtime used to reproduce the issue with controlled canary services only. It does not contain real secrets, real cloud metadata access, or third-party API keys.\n\n[PraisonAI-Runtime-Repro.zip](https://github.com/user-attachments/files/29142379/PraisonAI-Runtime-Repro.zip)\n\n## Impact\nSSRF against internal/loopback/cloud-metadata endpoints with **disclosure of internal HTTP responses** (read-back) to the attacker. Bypasses the project\u0027s SSRF protection on every fetch backend.\n\n## Suggested remediation\n1. Resolve the host once, reject all returned records that are private/loopback/link-local/ULA/CGNAT/metadata, then **connect to that exact validated IP** (pin it; send the original `Host`). Do not let the HTTP client / browser re-resolve.\n2. Apply the same validation + IP pinning to every backend (httpx, urllib fallback, crawl4ai) and every redirect hop.\n3. Disable automatic redirect following (or cap + re-validate each hop with pinning).\n4. Treat IPv4-mapped IPv6, decimal/octal/hex IPs, and CGNAT/non-global ranges as unsafe.",
"id": "GHSA-qg25-6gc4-48mg",
"modified": "2026-10-08T16:36:50Z",
"published": "2026-10-08T16:36:50Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-qg25-6gc4-48mg"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-61430"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-62169"
},
{
"type": "PACKAGE",
"url": "https://github.com/MervinPraison/PraisonAI"
},
{
"type": "WEB",
"url": "https://www.vulncheck.com/advisories/praisonai-before-dns-rebinding-ssrf-via-web-crawl"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:L/A:N",
"type": "CVSS_V3"
}
],
"summary": "PraisonAI: DNS rebinding bypass in `web_crawl` SSRF protection allows internal response disclosure"
}
Mitigation
The most basic advice for TOCTOU vulnerabilities is to not perform a check before the use. This does not resolve the underlying issue of the execution of a function on a resource whose state and identity cannot be assured, but it does help to limit the false sense of security given by the check.
Mitigation
When the file being altered is owned by the current user and group, set the effective gid and uid to that of the current user and group when executing this statement.
Mitigation
Limit the interleaving of operations on files from multiple processes.
Mitigation
If you cannot perform operations atomically and you must share access to the resource between multiple processes or threads, then try to limit the amount of time (CPU cycles) between the check and use of the resource. This will not fix the problem, but it could make it more difficult for an attack to succeed.
Mitigation
Recheck the resource after the use call to verify that the action was taken appropriately.
Mitigation
Ensure that some environmental locking mechanism can be used to protect resources effectively.
Mitigation
Ensure that locking occurs before the check, as opposed to afterwards, such that the resource, as checked, is the same as it is when in use.
CAPEC-27: Leveraging Race Conditions via Symbolic Links
This attack leverages the use of symbolic links (Symlinks) in order to write to sensitive files. An attacker can create a Symlink link to a target file not otherwise accessible to them. When the privileged program tries to create a temporary file with the same name as the Symlink link, it will actually write to the target file pointed to by the attackers' Symlink link. If the attacker can insert malicious content in the temporary file they will be writing to the sensitive file by using the Symlink. The race occurs because the system checks if the temporary file exists, then creates the file. The attacker would typically create the Symlink during the interval between the check and the creation of the temporary file.
CAPEC-29: Leveraging Time-of-Check and Time-of-Use (TOCTOU) Race Conditions
This attack targets a race condition occurring between the time of check (state) for a resource and the time of use of a resource. A typical example is file access. The adversary can leverage a file access race condition by "running the race", meaning that they would modify the resource between the first time the target program accesses the file and the time the target program uses the file. During that period of time, the adversary could replace or modify the file, causing the application to behave unexpectedly.