CWE-22
Weakness Allowed-with-ReviewImproper Limitation of a Pathname to a Restricted Directory
Abstraction: Base · Status: Stable
The product uses external input to construct a pathname that is intended to identify a file or directory that is located underneath a restricted parent directory, but the product does not properly neutralize special elements within the pathname that can cause the pathname to resolve to a location that is outside of the restricted directory.
Many file operations are intended to take place within a restricted directory. By using special elements such as ".." and "/" separators, attackers can escape outside of the restricted location to access files or directories that are elsewhere on the system. One of the most common special elements is the "../" sequence, which in most modern operating systems is interpreted as the parent directory of the current location. This is referred to as relative path traversal. Path traversal also covers the use of absolute pathnames such as "/usr/local/bin" to access unexpected files. This is referred to as absolute path traversal.
Mapping guidance
This CWE entry might have children that would be more appropriate.
Examine children of this entry to see if there is a better fit. Consider children such as CWE-23 (or its descendants) for relative path traversal, or CWE-36 for absolute path traversal.
Relationships and members
Modes of introduction
Detection methods
Automated Static Analysis
Automated techniques can find areas where path traversal weaknesses exist. However, tuning or customization may be required to remove or de-prioritize path-traversal problems that are only exploitable by the product's administrator - or other privileged users - and thus potentially valid behavior or, at worst, a bug instead of a vulnerability.
Effectiveness: High
Manual Static Analysis
Manual white box techniques may be able to provide sufficient code coverage and reduction of false positives if all file access operations can be assessed within limited time constraints.
Effectiveness: High
Automated Static Analysis - Binary or Bytecode
According to SOAR [REF-1479], the following detection techniques may be useful:
Highly cost effective:
Bytecode Weakness Analysis - including disassembler + source code weakness analysis
Cost effective for partial coverage:
Binary Weakness Analysis - including disassembler + source code weakness analysis
Effectiveness: High
Manual Static Analysis - Binary or Bytecode
According to SOAR [REF-1479], the following detection techniques may be useful:
Cost effective for partial coverage:
Binary / Bytecode disassembler - then use manual analysis for vulnerabilities & anomalies
Effectiveness: SOAR Partial
Dynamic Analysis with Automated Results Interpretation
According to SOAR [REF-1479], the following detection techniques may be useful:
Highly cost effective:
Web Application Scanner
Web Services Scanner
Database Scanners
Effectiveness: High
Dynamic Analysis with Manual Results Interpretation
According to SOAR [REF-1479], the following detection techniques may be useful:
Highly cost effective:
Fuzz Tester
Framework-based Fuzzer
Effectiveness: High
Manual Static Analysis - Source Code
According to SOAR [REF-1479], the following detection techniques may be useful:
Highly cost effective:
Manual Source Code Review (not inspections)
Cost effective for partial coverage:
Focused Manual Spotcheck - Focused manual analysis of source
Effectiveness: High
Automated Static Analysis - Source Code
According to SOAR [REF-1479], the following detection techniques may be useful:
Highly cost effective:
Source code Weakness Analyzer
Context-configured Source Code Weakness Analyzer
Effectiveness: High
Architecture or Design Review
According to SOAR [REF-1479], the following detection techniques may be useful:
Highly cost effective:
Formal Methods / Correct-By-Construction
Cost effective for partial coverage:
Inspection (IEEE 1028 standard) (can apply to requirements, design, source code, etc.)
Effectiveness: High
Notes
Other
In many programming languages, the injection of a null byte (the 0 or NUL) may allow an attacker to truncate a generated filename to apply to a wider range of files. For example, the product may add ".txt" to any pathname, thus limiting the attacker to text files, but a null injection may effectively remove this restriction.
Relationship
Pathname equivalence can be regarded as a type of canonicalization error.
Relationship
Some pathname equivalence issues are not directly related to directory traversal, rather are used to bypass security-relevant checks for whether a file/directory can be accessed by the attacker (e.g. a trailing "/" on a filename could bypass access rules that don't expect a trailing /, causing a server to provide the file when it normally would not).
Terminology
Like other weaknesses, terminology is often based on the types of manipulations used, instead of the underlying weaknesses. Some people use "directory traversal" only to refer to the injection of ".." and equivalent sequences whose specific meaning is to traverse directories.
Other variants like "absolute pathname" and "drive letter" have the *effect* of directory traversal, but some people may not call it such, since it doesn't involve ".." or equivalent.
Research Gap
Many variants of path traversal attacks are probably under-studied with respect to root cause. CWE-790 and CWE-182 begin to cover part of this gap.
Research Gap
Incomplete diagnosis or reporting of vulnerabilities can make it difficult to know which variant is affected. For example, a researcher might say that "..\" is vulnerable, but not test "../" which may also be vulnerable.
Any combination of directory separators ("/", "\", etc.) and numbers of "." (e.g. "....") can produce unique variants; for example, the "//../" variant is not listed (CVE-2004-0325). See this entry's children and lower-level descendants.
Content history
Identified weakness in Perl demonstrative example
Provided diagram to improve CWE usability
Identified weakness in "good code" for Python demonstrative example
Provided a demonstrative example in PHP
updated Potential_Mitigations, Time_of_Introduction
Suggested OWASP Top Ten 2004 mapping
updated Alternate_Terms, Relationships, Other_Notes, Relationship_Notes, Relevant_Properties, Taxonomy_Mappings, Weakness_Ordinalities
updated Description
updated Relationships, Taxonomy_Mappings
updated Potential_Mitigations
updated Alternate_Terms, Applicable_Platforms, Common_Consequences, Demonstrative_Examples, Description, Detection_Factors, Likelihood_of_Exploit, Name, Observed_Examples, Other_Notes, Potential_Mitigations, References, Related_Attack_Patterns, Relationship_Notes, Relationships, Research_Gaps, Taxonomy_Mappings, Terminology_Notes, Time_of_Introduction, Weakness_Ordinalities
updated Common_Consequences, Demonstrative_Examples, Description, Detection_Factors, Potential_Mitigations, References, Relationships
updated Potential_Mitigations
updated Potential_Mitigations
updated Potential_Mitigations
updated Relationships
updated Potential_Mitigations, References, Relationships, Taxonomy_Mappings
updated Demonstrative_Examples, References, Relationships
updated Potential_Mitigations
updated Observed_Examples
updated Related_Attack_Patterns, Relationships
updated Other_Notes, Research_Gaps
updated Detection_Factors, Relationships, Taxonomy_Mappings
updated Relationships
updated Related_Attack_Patterns
updated Demonstrative_Examples
updated Affected_Resources, Causal_Nature, Likelihood_of_Exploit, References, Relationships, Relevant_Properties, Taxonomy_Mappings
updated References, Relationships
updated References, Related_Attack_Patterns, Relationships, Taxonomy_Mappings
updated Related_Attack_Patterns, Relationships, Type
updated Relationships
updated Potential_Mitigations, Relationships
updated Demonstrative_Examples, Potential_Mitigations
updated Relationships
updated Potential_Mitigations, Relationships
updated Demonstrative_Examples, Relationships
updated Relationships
updated Observed_Examples, Relationships
updated Observed_Examples, Relationships
updated Observed_Examples, References
updated Common_Consequences, Description, Detection_Factors
updated Demonstrative_Examples, References, Relationships, Time_of_Introduction
updated Mapping_Notes, Relationships
updated Observed_Examples
updated Common_Consequences, Description, Diagram, Observed_Examples, Other_Notes, References
updated Demonstrative_Examples, Relationships
updated Relationships
updated Applicable_Platforms, Detection_Factors, Observed_Examples, Potential_Mitigations, References
updated Mapping_Notes, Relationships
updated Alternate_Terms, Applicable_Platforms, Demonstrative_Examples, Relationships
updated Applicable_Platforms, Common_Consequences, Name, References, Relationships, Taxonomy_Mappings
16162 vulnerabilities reference this CWE, most recent first.
GHSA-5RMH-7P7V-FMFC
Vulnerability from github – Published: 2022-12-19 15:30 – Updated: 2022-12-23 18:30The InPost Gallery WordPress plugin before 2.1.4.1 insecurely uses PHP's extract() function when rendering HTML views, allowing attackers to force the inclusion of malicious files & URLs, which may enable them to run code on servers.
{
"affected": [],
"aliases": [
"CVE-2022-4063"
],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2022-12-19T14:15:00Z",
"severity": "CRITICAL"
},
"details": "The InPost Gallery WordPress plugin before 2.1.4.1 insecurely uses PHP\u0027s extract() function when rendering HTML views, allowing attackers to force the inclusion of malicious files \u0026 URLs, which may enable them to run code on servers.",
"id": "GHSA-5rmh-7p7v-fmfc",
"modified": "2022-12-23T18:30:40Z",
"published": "2022-12-19T15:30:29Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2022-4063"
},
{
"type": "WEB",
"url": "https://wpscan.com/vulnerability/6bb07ec1-f1aa-4f4b-9717-c92f651a90a7"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:H",
"type": "CVSS_V3"
}
]
}
GHSA-5RMQ-CHC7-M22F
Vulnerability from github – Published: 2026-10-02 22:44 – Updated: 2026-10-02 22:44Summary:
2 findings — safe_user_path() accepts any path under Path.home() or Path.cwd(), which inside the shipped root container resolves to /root and /app (so all of root's home, including /root/.ssh/id_rsa, /root/.aws/credentials, /root/.kube/config, and /app/agent/.env, passes the check) (F9). read_document() has no sandbox call at all and returns the full content of any path the FastAPI process can read, including /etc/shadow, /etc/passwd, /proc/self/environ, and any secret file mounted into the container (F10). F10 is strictly broader than F9 but they have different fix scopes (F10 = a missing safe_path() call in one function; F9 = the envelope definition in path_utils.py), so both must be patched.
Shared baseline (applies to both findings)
The container has no USER directive (Dockerfile:15 — FROM python:3.11-slim AS runtime, no subsequent USER), so the FastAPI process runs as uid=0(root).
The two file-read tools described here are members of the auto-discovered LLM tool registry.
Combined with GHSA-1 / F1, they are reachable from any anonymous TCP client to port 8899, but the same defects also apply to authenticated sessions and to prompt-injection in any document the agent processes.
See GHSA-1's shared reproducer block for the install steps; the same docker compose up -d setup applies here.
Note on the
HOSTplaceholder used throughout the per-finding "Steps to observe" blocks below: replaceHOSTwith the address you reach the docker host on — typicallylocalhost(or127.0.0.1) if you are running the reproducer on the same machine as the container. Allcurlcommands below assume this substitution.
Finding 9 — High: safe_user_path() accepts the entire user home directory and process CWD, allowing LLM tool calls to read /root credentials
- Severity: High
- CVSS v3.1: 7.5 —
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N - CVSS v4.0: 8.7 —
AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N - CWE: CWE-22 (Path Traversal); CWE-552 (Files Accessible to External Parties)
Affected file: agent/src/tools/path_utils.py
- line 52 — def safe_user_path(p: str) -> Path:
- line 73-77 — if resolved.is_relative_to(home) or resolved.is_relative_to(cwd): return resolved — home = Path.home(), cwd = Path.cwd()
Intent vs actual:
safe_user_path() is intended to permit journal and shadow-account tools to open broker export files the operator may have placed anywhere under their home directory or the project folder. The intended invariant is that only user-owned broker data files are accessible — not system credential files or SSH keys. The actual envelope check accepts any path whose resolved form is inside Path.home() or Path.cwd(). Inside the shipped Docker container, Path.home() resolves to /root and Path.cwd() resolves to /app. Every file under either subtree passes the check, including:
/root/.ssh/id_rsaand any other SSH key files/root/.aws/credentials,/root/.kube/config,/root/.docker/config.json/app/agent/.env(the file containing the operator's realOPENROUTER_API_KEY,TUSHARE_TOKEN, and any other secrets)
A runtime probe inside the container confirmed that safe_user_path('/root/.aws/credentials') returned the path without raising ValueError. ExtractShadowStrategyTool was then invoked against /root/secrets/aws.csv (a planted credential file) and returned an error message containing the first line of the file via the parse-error channel.
Steps to observe:
- Per GHSA-1 shared reproducer, start the server with a working LLM API key and create an unauthenticated session.
- (Setup for safe demo: inside the container,
docker execa planted file:docker exec <container> sh -c 'mkdir -p /root/secrets && printf "broker_id,api_key,api_secret\nDEMO,FAKE_KEY,FAKE_SECRET\n" > /root/secrets/aws.csv'.) curl -s -X POST "http://HOST:8899/sessions/$SID/messages" -H 'Content-Type: application/json' -d '{"content":"Analyze the trade journal at the path /root/secrets/aws.csv and tell me what you find."}'- Poll
curl -s "http://HOST:8899/sessions/$SID/messages". Observe the agent invokeExtractShadowStrategyToolwithjournal_path="/root/secrets/aws.csv", which passessafe_user_path()and attempts to parse the file as a trade journal CSV. - Observe the error response — when the file's structure does not match the expected journal schema, the parse error often includes the first line (column names) verbatim, leaking the file's first line.
- Repeat with
journal_path="/app/agent/.env"to confirm the.envfile is within the accepted envelope.
Impact:
Any unauthenticated caller can instruct the LLM to attempt to parse any file under /root or /app as a trade journal, extracting the file's first line via the parse-error message channel. Files with valid CSV-like first lines may leak multiple bytes. In the shipped root container, /root encompasses all credentials a careless operator may have mounted into the home directory; /app includes the agent's own secrets and any operator-staged data files.
Finding 10 — High: read_document() opens any server-readable file with no sandbox enforcement, returning full content of /etc/shadow and /proc/self/environ
- Severity: High
- CVSS v3.1: 7.5 —
AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N - CVSS v4.0: 8.7 —
AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N - CWE: CWE-22 (Path Traversal); CWE-552 (Files Accessible to External Parties); CWE-200 (Information Exposure)
Affected file: agent/src/tools/doc_reader_tool.py
- line 259 — def read_document(file_path: str, pages: str = "") -> str:
- line 270 — path = Path(file_path) — followed only by path.exists() and path.is_file() checks before dispatching to format-specific readers
- No call to safe_path, safe_user_path, or any other sandbox enforcement appears anywhere in the function
Intent vs actual: DocReaderTool is intended to allow the LLM agent to read documents and data files provided for analysis. Like other file-reading tools in the project, it should apply a sandbox check before opening the file. The actual implementation takes the LLM-emitted file_path string, runs only path.exists() and path.is_file(), and dispatches to the appropriate reader. No call to safe_path or safe_user_path exists in the function. A runtime probe confirmed:
read_document('/etc/passwd')returned HTTP 200 with 839 characters of contentread_document('/etc/shadow')returned the full shadow password fileread_document('/proc/self/environ')returned the full process environment, includingOPENROUTER_API_KEYandTUSHARE_TOKENin plaintext
This is strictly wider than F9: F9 is bounded to /root + /app via the (overly-broad) envelope; F10 has no envelope at all and reaches /etc, /proc, /var, and any other path the FastAPI process can read.
Steps to observe:
- Per GHSA-1 shared reproducer, start the server with a working LLM API key and create an unauthenticated session.
curl -s -X POST "http://HOST:8899/sessions/$SID/messages" -H 'Content-Type: application/json' -d '{"content":"Please read and summarize the document at /proc/self/environ"}'- Poll
curl -s "http://HOST:8899/sessions/$SID/messages". Observe the agent invokeread_documentwithfile_path="/proc/self/environ"and return the full process environment in the message stream. - Observe
OPENROUTER_API_KEY,TUSHARE_TOKEN, and any other variables inagent/.envappearing in plaintext. - Repeat with
file_path="/etc/shadow"to confirm shadow password file access.
Impact:
An unauthenticated caller can retrieve any file the server process can read. Running as root, that includes /etc/shadow, /etc/passwd, /proc/self/environ (full plaintext API keys), /root/.ssh/id_rsa, and any secret files mounted into the container. This is the broadest file-read primitive in the codebase and provides a credential-extraction path that does not require shell execution — endpoint monitoring tuned to BashTool / shell signatures will miss it entirely.
Why F9 and F10 are listed separately
A maintainer might be tempted to fix only one, on the theory that F10 dominates F9. Two reasons to fix both:
-
Different fix scope — F10's fix is a single missing call (
safe_path(file_path)inread_documentbefore line 270). F9's fix is insafe_user_path()itself: the envelope must be replaced with a strict allowlist of operator-configured directories, notPath.home() ∪ Path.cwd(). A fix that adds the missingsafe_user_pathcall toread_documentis insufficient becausesafe_user_pathitself accepts/rootand/app/agent/.env. Both surfaces need work. -
Different reachability classes — F9 is reachable through tools that already gate on
safe_user_path(ExtractShadowStrategyTooland several journal tools), so even a hypothetical F10 fix that switchedread_documentto usesafe_user_pathwould still leak/root/*because the envelope is broken. F9 is the structural defect; F10 is the missed call.
Suggested remediation
-
F9 — In
safe_user_path()atpath_utils.py:52-77, replace thePath.home() ∪ Path.cwd()envelope with a strict allowlist of operator-configured directories (e.g. an explicitBROKER_EXPORTS_DIRenv var defaulting to/app/data/broker_exports/). Reject/root,/app/agent/.env, and/app/agent/uploads/(the latter to prevent F3-uploaded files from being subsequently parsed as a credential-leak vector via the parse-error channel).' -
F10 — Add a
safe_path()(orsafe_user_path()) call atdoc_reader_tool.py:270before the existingpath.exists()/path.is_file()checks. Once F9 is patched, the same allowlist will apply uniformly to bothread_documentand thesafe_user_path-gated tools. -
Defense-in-depth — Drop the FastAPI process to a non-root user. Add a
RUN useradd -m vibe && chown -R vibe /appstep to the Dockerfile andUSER vibebeforeCMD. This does not fix the Path Traversal but materially reduces the credential-extraction blast radius of any successful exploit (and benefits every other finding in GHSA-1 and GHSA-2). See GHSA-1 / shared baseline for the matchingUSERrecommendation.
{
"affected": [
{
"package": {
"ecosystem": "PyPI",
"name": "vibe-trading-ai"
},
"ranges": [
{
"events": [
{
"introduced": "0.1.0"
},
{
"fixed": "0.1.7"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [],
"database_specific": {
"cwe_ids": [
"CWE-200",
"CWE-22",
"CWE-23",
"CWE-552"
],
"github_reviewed": true,
"github_reviewed_at": "2026-10-02T22:44:12Z",
"nvd_published_at": null,
"severity": "HIGH"
},
"details": "### Summary: \n2 findings \u2014 `safe_user_path()` accepts any path under `Path.home()` or `Path.cwd()`, which inside the shipped root container resolves to `/root` and `/app` (so all of root\u0027s home, including `/root/.ssh/id_rsa`, `/root/.aws/credentials`, `/root/.kube/config`, and `/app/agent/.env`, passes the check) (F9). `read_document()` has no sandbox call at all and returns the full content of any path the FastAPI process can read, including `/etc/shadow`, `/etc/passwd`, `/proc/self/environ`, and any secret file mounted into the container (F10). F10 is strictly broader than F9 but they have different fix scopes (F10 = a missing `safe_path()` call in one function; F9 = the envelope definition in `path_utils.py`), so both must be patched.\n\n---\n\n### Shared baseline (applies to both findings)\n\nThe container has no `USER` directive (Dockerfile:15 \u2014 `FROM python:3.11-slim AS runtime`, no subsequent `USER`), so the FastAPI process runs as `uid=0(root)`. \n\nThe two file-read tools described here are members of the auto-discovered LLM tool registry. \nCombined with GHSA-1 / F1, they are reachable from any anonymous TCP client to port 8899, but the same defects also apply to authenticated sessions and to prompt-injection in any document the agent processes. \nSee GHSA-1\u0027s shared reproducer block for the install steps; the same `docker compose up -d` setup applies here.\n\n\u003e **Note on the `HOST` placeholder used throughout the per-finding \"Steps to observe\" blocks below**: replace `HOST` with the address you reach the docker host on \u2014 typically `localhost` (or `127.0.0.1`) if you are running the reproducer on the same machine as the container. All `curl` commands below assume this substitution.\n\n---\n\n### Finding 9 \u2014 High: safe_user_path() accepts the entire user home directory and process CWD, allowing LLM tool calls to read /root credentials\n\n- **Severity**: High\n- **CVSS v3.1**: 7.5 \u2014 `AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N`\n- **CVSS v4.0**: 8.7 \u2014 `AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N`\n- **CWE**: CWE-22 (Path Traversal); CWE-552 (Files Accessible to External Parties)\n\n**Affected file**: `agent/src/tools/path_utils.py`\n- line 52 \u2014 `def safe_user_path(p: str) -\u003e Path:`\n- line 73-77 \u2014 `if resolved.is_relative_to(home) or resolved.is_relative_to(cwd): return resolved` \u2014 `home = Path.home()`, `cwd = Path.cwd()`\n\n**Intent vs actual**: \n`safe_user_path()` is intended to permit journal and shadow-account tools to open broker export files the operator may have placed anywhere under their home directory or the project folder. The intended invariant is that only user-owned broker data files are accessible \u2014 not system credential files or SSH keys. The actual envelope check accepts any path whose resolved form is inside `Path.home()` or `Path.cwd()`. Inside the shipped Docker container, `Path.home()` resolves to `/root` and `Path.cwd()` resolves to `/app`. Every file under either subtree passes the check, including:\n\n- `/root/.ssh/id_rsa` and any other SSH key files\n- `/root/.aws/credentials`, `/root/.kube/config`, `/root/.docker/config.json`\n- `/app/agent/.env` (the file containing the operator\u0027s real `OPENROUTER_API_KEY`, `TUSHARE_TOKEN`, and any other secrets)\n\nA runtime probe inside the container confirmed that `safe_user_path(\u0027/root/.aws/credentials\u0027)` returned the path without raising `ValueError`. `ExtractShadowStrategyTool` was then invoked against `/root/secrets/aws.csv` (a planted credential file) and returned an error message containing the first line of the file via the parse-error channel.\n\n**Steps to observe**:\n\n1. Per GHSA-1 shared reproducer, start the server with a working LLM API key and create an unauthenticated session.\n2. (Setup for safe demo: inside the container, `docker exec` a planted file: `docker exec \u003ccontainer\u003e sh -c \u0027mkdir -p /root/secrets \u0026\u0026 printf \"broker_id,api_key,api_secret\\nDEMO,FAKE_KEY,FAKE_SECRET\\n\" \u003e /root/secrets/aws.csv\u0027`.)\n3. `curl -s -X POST \"http://HOST:8899/sessions/$SID/messages\" -H \u0027Content-Type: application/json\u0027 -d \u0027{\"content\":\"Analyze the trade journal at the path /root/secrets/aws.csv and tell me what you find.\"}\u0027`\n4. Poll `curl -s \"http://HOST:8899/sessions/$SID/messages\"`. Observe the agent invoke `ExtractShadowStrategyTool` with `journal_path=\"/root/secrets/aws.csv\"`, which passes `safe_user_path()` and attempts to parse the file as a trade journal CSV.\n5. Observe the error response \u2014 when the file\u0027s structure does not match the expected journal schema, the parse error often includes the first line (column names) verbatim, leaking the file\u0027s first line.\n6. Repeat with `journal_path=\"/app/agent/.env\"` to confirm the `.env` file is within the accepted envelope.\n\n**Impact**: \nAny unauthenticated caller can instruct the LLM to attempt to parse any file under `/root` or `/app` as a trade journal, extracting the file\u0027s first line via the parse-error message channel. Files with valid CSV-like first lines may leak multiple bytes. In the shipped root container, `/root` encompasses all credentials a careless operator may have mounted into the home directory; `/app` includes the agent\u0027s own secrets and any operator-staged data files.\n\n---\n\n### Finding 10 \u2014 High: read_document() opens any server-readable file with no sandbox enforcement, returning full content of /etc/shadow and /proc/self/environ\n\n- **Severity**: High\n- **CVSS v3.1**: 7.5 \u2014 `AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N`\n- **CVSS v4.0**: 8.7 \u2014 `AV:N/AC:L/AT:N/PR:N/UI:N/VC:H/VI:N/VA:N/SC:N/SI:N/SA:N`\n- **CWE**: CWE-22 (Path Traversal); CWE-552 (Files Accessible to External Parties); CWE-200 (Information Exposure)\n\n**Affected file**: `agent/src/tools/doc_reader_tool.py`\n- line 259 \u2014 `def read_document(file_path: str, pages: str = \"\") -\u003e str:`\n- line 270 \u2014 `path = Path(file_path)` \u2014 followed only by `path.exists()` and `path.is_file()` checks before dispatching to format-specific readers\n- No call to `safe_path`, `safe_user_path`, or any other sandbox enforcement appears anywhere in the function\n\n**Intent vs actual**: `DocReaderTool` is intended to allow the LLM agent to read documents and data files provided for analysis. Like other file-reading tools in the project, it should apply a sandbox check before opening the file. The actual implementation takes the LLM-emitted `file_path` string, runs only `path.exists()` and `path.is_file()`, and dispatches to the appropriate reader. No call to `safe_path` or `safe_user_path` exists in the function. A runtime probe confirmed:\n\n- `read_document(\u0027/etc/passwd\u0027)` returned HTTP 200 with 839 characters of content\n- `read_document(\u0027/etc/shadow\u0027)` returned the full shadow password file\n- `read_document(\u0027/proc/self/environ\u0027)` returned the full process environment, including `OPENROUTER_API_KEY` and `TUSHARE_TOKEN` in plaintext\n\nThis is strictly **wider** than F9: F9 is bounded to `/root` + `/app` via the (overly-broad) envelope; F10 has no envelope at all and reaches `/etc`, `/proc`, `/var`, and any other path the FastAPI process can read.\n\n**Steps to observe**:\n\n1. Per GHSA-1 shared reproducer, start the server with a working LLM API key and create an unauthenticated session.\n2. `curl -s -X POST \"http://HOST:8899/sessions/$SID/messages\" -H \u0027Content-Type: application/json\u0027 -d \u0027{\"content\":\"Please read and summarize the document at /proc/self/environ\"}\u0027`\n3. Poll `curl -s \"http://HOST:8899/sessions/$SID/messages\"`. Observe the agent invoke `read_document` with `file_path=\"/proc/self/environ\"` and return the full process environment in the message stream.\n4. Observe `OPENROUTER_API_KEY`, `TUSHARE_TOKEN`, and any other variables in `agent/.env` appearing in plaintext.\n5. Repeat with `file_path=\"/etc/shadow\"` to confirm shadow password file access.\n\n**Impact**: \nAn unauthenticated caller can retrieve any file the server process can read. Running as root, that includes `/etc/shadow`, `/etc/passwd`, `/proc/self/environ` (full plaintext API keys), `/root/.ssh/id_rsa`, and any secret files mounted into the container. This is the broadest file-read primitive in the codebase and provides a credential-extraction path that does **not** require shell execution \u2014 endpoint monitoring tuned to BashTool / shell signatures will miss it entirely.\n\n---\n\n### Why F9 and F10 are listed separately\n\nA maintainer might be tempted to fix only one, on the theory that F10 dominates F9. Two reasons to fix both:\n\n1. **Different fix scope** \u2014 F10\u0027s fix is a single missing call (`safe_path(file_path)` in `read_document` before line 270). F9\u0027s fix is in `safe_user_path()` itself: the envelope must be replaced with a strict allowlist of operator-configured directories, *not* `Path.home() \u222a Path.cwd()`. A fix that adds the missing `safe_user_path` call to `read_document` is **insufficient** because `safe_user_path` itself accepts `/root` and `/app/agent/.env`. Both surfaces need work.\n\n2. **Different reachability classes** \u2014 F9 is reachable through tools that already gate on `safe_user_path` (`ExtractShadowStrategyTool` and several journal tools), so even a hypothetical F10 fix that switched `read_document` to use `safe_user_path` would still leak `/root/*` because the envelope is broken. F9 is the structural defect; F10 is the missed call.\n\n---\n\n### Suggested remediation\n\n11. **F9** \u2014 In `safe_user_path()` at `path_utils.py:52-77`, replace the `Path.home() \u222a Path.cwd()` envelope with a strict allowlist of operator-configured directories (e.g. an explicit `BROKER_EXPORTS_DIR` env var defaulting to `/app/data/broker_exports/`). Reject `/root`, `/app/agent/.env`, and `/app/agent/uploads/` (the latter to prevent F3-uploaded files from being subsequently parsed as a credential-leak vector via the parse-error channel).\u0027\n\n12. **F10** \u2014 Add a `safe_path()` (or `safe_user_path()`) call at `doc_reader_tool.py:270` before the existing `path.exists()` / `path.is_file()` checks. Once F9 is patched, the same allowlist will apply uniformly to both `read_document` and the `safe_user_path`-gated tools.\n\n13. **Defense-in-depth** \u2014 Drop the FastAPI process to a non-root user. Add a `RUN useradd -m vibe \u0026\u0026 chown -R vibe /app` step to the Dockerfile and `USER vibe` before `CMD`. This does not fix the Path Traversal but materially reduces the credential-extraction blast radius of any successful exploit (and benefits every other finding in GHSA-1 and GHSA-2). See GHSA-1 / shared baseline for the matching `USER` recommendation.\n\n---",
"id": "GHSA-5rmq-chc7-m22f",
"modified": "2026-10-02T22:44:12Z",
"published": "2026-10-02T22:44:12Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/HKUDS/Vibe-Trading/security/advisories/GHSA-5rmq-chc7-m22f"
},
{
"type": "WEB",
"url": "https://github.com/HKUDS/Vibe-Trading/commit/9454d4a27a763b80e1d6eb5763b86c88e9e4e714"
},
{
"type": "PACKAGE",
"url": "https://github.com/HKUDS/Vibe-Trading"
},
{
"type": "WEB",
"url": "https://github.com/HKUDS/Vibe-Trading/releases/tag/v0.1.7"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N",
"type": "CVSS_V3"
}
],
"summary": "Vibe-Trading file-read tools expose arbitrary server-readable files"
}
GHSA-5RMR-HV98-XF8X
Vulnerability from github – Published: 2026-08-20 15:34 – Updated: 2026-08-20 15:34A path traversal vulnerability in ATutor allows an authenticated user to access files from other course directories when the AT_FORCE_GET_FILE configuration option is enabled. This can lead to unauthorized access to files and disclosure of information about the filesystem structure.
Product is no longer actively supported and the vulnerabilities have not been fixed. Only version 2.2.4 was tested and confirmed as vulnerable, other versions were not tested but might also be vulnerable.
{
"affected": [],
"aliases": [
"CVE-2026-64963"
],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-08-20T14:17:44Z",
"severity": "LOW"
},
"details": "A path traversal vulnerability in ATutor allows an authenticated user to access files from other course directories when the\u00a0AT_FORCE_GET_FILE\u00a0configuration option is enabled. This can lead to unauthorized access to files and disclosure of information about the filesystem structure.\n\n\n\n\n\n\n\n\nProduct is no longer actively supported and the vulnerabilities have not been fixed. Only version 2.2.4 was tested and confirmed as vulnerable, other versions were not tested but might also be vulnerable.",
"id": "GHSA-5rmr-hv98-xf8x",
"modified": "2026-08-20T15:34:17Z",
"published": "2026-08-20T15:34:17Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-64963"
},
{
"type": "WEB",
"url": "https://atutor.github.io"
},
{
"type": "WEB",
"url": "https://cert.pl/en/posts/2026/08/CVE-2026-64960"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:4.0/AV:N/AC:L/AT:P/PR:L/UI:N/VC:L/VI:N/VA:N/SC:L/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-5RQ4-664W-9X2C
Vulnerability from github – Published: 2026-02-25 22:34 – Updated: 2026-02-25 22:34The basic-ftp library contains a path traversal vulnerability in the downloadToDir() method. A malicious FTP server can send directory listings with filenames containing path traversal sequences (../) that cause files to be written outside the intended download directory.
Source-to-Sink Flow
1. SOURCE: FTP server sends LIST response
└─> "-rw-r--r-- 1 user group 1024 Jan 20 12:00 ../../../etc/passwd"
2. PARSER: parseListUnix.ts:100 extracts filename
└─> file.name = "../../../etc/passwd"
3. VALIDATION: parseListUnix.ts:101 checks
└─> if (name === "." || name === "..") ❌ (only filters exact matches)
└─> "../../../etc/passwd" !== "." && !== ".." ✅ PASSES
4. SINK: Client.ts:707 uses filename directly
└─> const localPath = join(localDirPath, file.name)
└─> join("/safe/download", "../../../etc/passwd")
└─> Result: "/safe/download/../../../etc/passwd" → resolves to "/etc/passwd"
5. FILE WRITE: Client.ts:512 opens file
└─> fsOpen(localPath, "w") → writes to /etc/passwd (outside intended directory)
Vulnerable Code
File: src/Client.ts:707
protected async _downloadFromWorkingDir(localDirPath: string): Promise<void> {
await ensureLocalDirectory(localDirPath)
for (const file of await this.list()) {
const localPath = join(localDirPath, file.name) // ⚠️ VULNERABLE
// file.name comes from untrusted FTP server, no sanitization
await this.downloadTo(localPath, file.name)
}
}
Root Cause:
- Parser validation (parseListUnix.ts:101) only filters exact . or .. entries
- No sanitization of ../ sequences in filenames
- path.join() doesn't prevent traversal, fs.open() resolves paths
Impact
A malicious FTP server can: - Write files to arbitrary locations on the client filesystem - Overwrite critical system files (if user has write access) - Potentially achieve remote code execution
Affected Versions
- Tested: v5.1.0
- Likely: All versions (code pattern exists since initial implementation)
Mitigation
Workaround: Do not use downloadToDir() with untrusted FTP servers.
Fix: Sanitize filenames before use:
import { basename } from 'path'
// In _downloadFromWorkingDir:
const sanitizedName = basename(file.name) // Strip path components
const localPath = join(localDirPath, sanitizedName)
{
"affected": [
{
"package": {
"ecosystem": "npm",
"name": "basic-ftp"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "5.2.0"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"aliases": [
"CVE-2026-27699"
],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": true,
"github_reviewed_at": "2026-02-25T22:34:26Z",
"nvd_published_at": "2026-02-25T15:20:53Z",
"severity": "CRITICAL"
},
"details": "The `basic-ftp`\u00a0library contains a path traversal vulnerability in the `downloadToDir()`\u00a0method. A malicious FTP server can send directory listings with filenames containing path traversal sequences (`../`) that cause files to be written outside the intended download directory.\n\n\n## Source-to-Sink Flow\n\n```\n1. SOURCE: FTP server sends LIST response\n\u2514\u2500\u003e \"-rw-r--r-- 1 user group 1024 Jan 20 12:00 ../../../etc/passwd\"\n\n2. PARSER: parseListUnix.ts:100 extracts filename\n\u2514\u2500\u003e file.name = \"../../../etc/passwd\"\n\n3. VALIDATION: parseListUnix.ts:101 checks\n\u2514\u2500\u003e if (name === \".\" || name === \"..\") \u274c (only filters exact matches)\n\u2514\u2500\u003e \"../../../etc/passwd\" !== \".\" \u0026\u0026 !== \"..\" \u2705 PASSES\n\n4. SINK: Client.ts:707 uses filename directly\n\u2514\u2500\u003e const localPath = join(localDirPath, file.name)\n\u2514\u2500\u003e join(\"/safe/download\", \"../../../etc/passwd\")\n\u2514\u2500\u003e Result: \"/safe/download/../../../etc/passwd\" \u2192 resolves to \"/etc/passwd\"\n\n5. FILE WRITE: Client.ts:512 opens file\n\u2514\u2500\u003e fsOpen(localPath, \"w\") \u2192 writes to /etc/passwd (outside intended directory)\n```\n\n## Vulnerable Code\n\n**File**: `src/Client.ts:707`\n\n```typescript\nprotected async _downloadFromWorkingDir(localDirPath: string): Promise\u003cvoid\u003e\u00a0{\nawait ensureLocalDirectory(localDirPath)\nfor\u00a0(const file of await\u00a0this.list()) {\nconst localPath = join(localDirPath, file.name) // \u26a0\ufe0f VULNERABLE\n// file.name comes from untrusted FTP server, no sanitization\nawait this.downloadTo(localPath, file.name)\n}\n}\n```\n\n**Root Cause**:\n-\u00a0Parser validation (`parseListUnix.ts:101`) only filters exact `.`\u00a0or `..`\u00a0entries\n-\u00a0No sanitization of `../`\u00a0sequences in filenames\n- `path.join()`\u00a0doesn\u0027t prevent traversal, `fs.open()`\u00a0resolves paths\n\n\n# Impact\n\nA malicious FTP server can:\n-\u00a0Write files to arbitrary locations on the client filesystem\n-\u00a0Overwrite critical system files (if user has write access)\n-\u00a0Potentially achieve remote code execution\n\n## Affected Versions\n\n- **Tested**: v5.1.0\n- **Likely**: All versions (code pattern exists since initial implementation)\n\n## Mitigation\n\n**Workaround**: Do not use `downloadToDir()`\u00a0with untrusted FTP servers.\n\n**Fix**: Sanitize filenames before use:\n\n```typescript\nimport\u00a0{ basename\u00a0} from \u0027path\u0027\n\n// In _downloadFromWorkingDir:\nconst sanitizedName = basename(file.name) // Strip path components\nconst localPath = join(localDirPath, sanitizedName)\n```",
"id": "GHSA-5rq4-664w-9x2c",
"modified": "2026-02-25T22:34:26Z",
"published": "2026-02-25T22:34:26Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/patrickjuchli/basic-ftp/security/advisories/GHSA-5rq4-664w-9x2c"
},
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-27699"
},
{
"type": "WEB",
"url": "https://github.com/patrickjuchli/basic-ftp/commit/2a2a0e6514357b9eda07c2f8afbd3f04727a7cd9"
},
{
"type": "PACKAGE",
"url": "https://github.com/patrickjuchli/basic-ftp"
},
{
"type": "WEB",
"url": "https://github.com/patrickjuchli/basic-ftp/releases/tag/v5.2.0"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:H",
"type": "CVSS_V3"
}
],
"summary": "Basic FTP has Path Traversal Vulnerability in its downloadToDir()\u00a0method"
}
GHSA-5RQJ-G24X-G4WV
Vulnerability from github – Published: 2023-11-29 06:30 – Updated: 2023-12-05 15:30Dreamer CMS before version 4.0.1 is vulnerable to Directory Traversal. Background template management allows arbitrary modification of the template file, allowing system sensitive files to be read.
{
"affected": [],
"aliases": [
"CVE-2023-46886"
],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2023-11-29T05:15:07Z",
"severity": "CRITICAL"
},
"details": "Dreamer CMS before version 4.0.1 is vulnerable to Directory Traversal. Background template management allows arbitrary modification of the template file, allowing system sensitive files to be read.",
"id": "GHSA-5rqj-g24x-g4wv",
"modified": "2023-12-05T15:30:34Z",
"published": "2023-11-29T06:30:22Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2023-46886"
},
{
"type": "WEB",
"url": "https://gitee.com/iteachyou/dreamer_cms/issues/I6NOFN"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:H/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-5RV2-FPHG-FGCG
Vulnerability from github – Published: 2022-05-24 17:45 – Updated: 2022-05-24 17:45A vulnerability in the web UI of Cisco IOS XE Software could allow an authenticated, remote attacker to inject arbitrary commands that can be executed as the root user. This vulnerability is due to insufficient input validation. An attacker could exploit this vulnerability by sending a crafted request to the web UI of an affected device with arbitrary commands injected into a portion of the request. A successful exploit could allow the attacker to execute arbitrary commands as the root user.
{
"affected": [],
"aliases": [
"CVE-2021-1435"
],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2021-03-24T20:15:00Z",
"severity": "HIGH"
},
"details": "A vulnerability in the web UI of Cisco IOS XE Software could allow an authenticated, remote attacker to inject arbitrary commands that can be executed as the root user. This vulnerability is due to insufficient input validation. An attacker could exploit this vulnerability by sending a crafted request to the web UI of an affected device with arbitrary commands injected into a portion of the request. A successful exploit could allow the attacker to execute arbitrary commands as the root user.",
"id": "GHSA-5rv2-fphg-fgcg",
"modified": "2022-05-24T17:45:14Z",
"published": "2022-05-24T17:45:14Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2021-1435"
},
{
"type": "WEB",
"url": "https://tools.cisco.com/security/center/content/CiscoSecurityAdvisory/cisco-sa-iosxe-webcmdinjsh-UFJxTgZD"
}
],
"schema_version": "1.4.0",
"severity": []
}
GHSA-5RV5-FHMR-72HJ
Vulnerability from github – Published: 2022-05-12 00:01 – Updated: 2022-05-21 00:00ftcms <=2.1 was discovered to be vulnerable to directory traversal attacks via the parameter tp.
{
"affected": [],
"aliases": [
"CVE-2022-30061"
],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2022-05-11T18:15:00Z",
"severity": "MODERATE"
},
"details": "ftcms \u003c=2.1 was discovered to be vulnerable to directory traversal attacks via the parameter tp.",
"id": "GHSA-5rv5-fhmr-72hj",
"modified": "2022-05-21T00:00:49Z",
"published": "2022-05-12T00:01:00Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2022-30061"
},
{
"type": "WEB",
"url": "https://www.yuque.com/docs/share/bb31a8e0-0f86-4edf-bdea-3b4e7162e344"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:N/A:N",
"type": "CVSS_V3"
}
]
}
GHSA-5RVH-FH5R-RHFV
Vulnerability from github – Published: 2022-05-17 03:46 – Updated: 2022-05-17 03:46Multiple directory traversal vulnerabilities in GNU C Library (aka glibc or libc6) before 2.20 allow context-dependent attackers to bypass ForceCommand restrictions and possibly have other unspecified impact via a .. (dot dot) in a (1) LC_*, (2) LANG, or other locale environment variable.
{
"affected": [],
"aliases": [
"CVE-2014-0475"
],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2014-07-29T14:55:00Z",
"severity": "MODERATE"
},
"details": "Multiple directory traversal vulnerabilities in GNU C Library (aka glibc or libc6) before 2.20 allow context-dependent attackers to bypass ForceCommand restrictions and possibly have other unspecified impact via a .. (dot dot) in a (1) LC_*, (2) LANG, or other locale environment variable.",
"id": "GHSA-5rvh-fh5r-rhfv",
"modified": "2022-05-17T03:46:13Z",
"published": "2022-05-17T03:46:13Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2014-0475"
},
{
"type": "WEB",
"url": "https://rhn.redhat.com/errata/RHSA-2014-1110.html"
},
{
"type": "WEB",
"url": "https://security.gentoo.org/glsa/201602-02"
},
{
"type": "WEB",
"url": "https://sourceware.org/bugzilla/show_bug.cgi?id=17137"
},
{
"type": "WEB",
"url": "http://linux.oracle.com/errata/ELSA-2015-0092.html"
},
{
"type": "WEB",
"url": "http://www.debian.org/security/2014/dsa-2976"
},
{
"type": "WEB",
"url": "http://www.mandriva.com/security/advisories?name=MDVSA-2014:152"
},
{
"type": "WEB",
"url": "http://www.openwall.com/lists/oss-security/2014/07/10/7"
},
{
"type": "WEB",
"url": "http://www.openwall.com/lists/oss-security/2014/07/14/6"
},
{
"type": "WEB",
"url": "http://www.securityfocus.com/bid/68505"
},
{
"type": "WEB",
"url": "http://www.securitytracker.com/id/1030569"
}
],
"schema_version": "1.4.0",
"severity": []
}
GHSA-5RVR-H867-J9W7
Vulnerability from github – Published: 2022-05-24 19:09 – Updated: 2022-05-24 19:09A path traversal in Thrift command handlers in IDEMIA Morpho Wave Compact and VisionPass devices before 2.6.2 allows remote authenticated attackers to achieve denial of services and information disclosure via TCP/IP packets.
{
"affected": [],
"aliases": [
"CVE-2021-35521"
],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2021-07-22T12:15:00Z",
"severity": "MODERATE"
},
"details": "A path traversal in Thrift command handlers in IDEMIA Morpho Wave Compact and VisionPass devices before 2.6.2 allows remote authenticated attackers to achieve denial of services and information disclosure via TCP/IP packets.",
"id": "GHSA-5rvr-h867-j9w7",
"modified": "2022-05-24T19:09:01Z",
"published": "2022-05-24T19:09:01Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2021-35521"
},
{
"type": "WEB",
"url": "https://biometricdevices.idemia.com/s/global-search/0696700000JJa0zAAD?sharing=true"
},
{
"type": "WEB",
"url": "https://biometricdevices.idemia.com/s/global-search/0696700000JJa1nAAD?sharing=true"
},
{
"type": "WEB",
"url": "https://www.idemia.com"
}
],
"schema_version": "1.4.0",
"severity": []
}
GHSA-5RWQ-2VWH-7G73
Vulnerability from github – Published: 2026-10-02 00:31 – Updated: 2026-10-06 15:31langflow-ai langflow v1.8.4 is affected by: Directory Traversal. The impact is: Arbitrary file write outside the intended workspace or storage boundary.. The component is: src/backend/base/langflow/api/v1/knowledge_bases.py:knowledge_bases-create_knowledge_base-a-live-http-post-to-create-knowledge-base. The attack vector is: Attack surface: HTTP or browser-backed service path. A public-facing upload or HTTP route handler forwards an attacker-controlled path or filename into host file creation without any visible boundary enforcement. ¶¶ A weakness has been identified in langflow-ai langflow up to 1.8.4. langflow contains an absolute path traversal vulnerability in knowledge_bases-create_knowledge_base-a-live-http-post-to-create-knowledge-base (src/backend/base/langflow/api/v1/knowledge_bases.py:51). An attacker can write or overwrite files outside the intended working directory by providing absolute paths in the knowledge base creation endpoint.
{
"affected": [],
"aliases": [
"CVE-2026-51888"
],
"database_specific": {
"cwe_ids": [
"CWE-22"
],
"github_reviewed": false,
"github_reviewed_at": null,
"nvd_published_at": "2026-10-01T22:17:03Z",
"severity": "HIGH"
},
"details": "langflow-ai langflow v1.8.4 is affected by: Directory Traversal. The impact is: Arbitrary file write outside the intended workspace or storage boundary.. The component is: src/backend/base/langflow/api/v1/knowledge_bases.py:knowledge_bases-create_knowledge_base-a-live-http-post-to-create-knowledge-base. The attack vector is: Attack surface: HTTP or browser-backed service path. A public-facing upload or HTTP route handler forwards an attacker-controlled path or filename into host file creation without any visible boundary enforcement. \u00b6\u00b6 A weakness has been identified in langflow-ai langflow up to 1.8.4. langflow contains an absolute path traversal vulnerability in knowledge_bases-create_knowledge_base-a-live-http-post-to-create-knowledge-base (src/backend/base/langflow/api/v1/knowledge_bases.py:51). An attacker can write or overwrite files outside the intended working directory by providing absolute paths in the knowledge base creation endpoint.",
"id": "GHSA-5rwq-2vwh-7g73",
"modified": "2026-10-06T15:31:28Z",
"published": "2026-10-02T00:31:30Z",
"references": [
{
"type": "ADVISORY",
"url": "https://nvd.nist.gov/vuln/detail/CVE-2026-51888"
},
{
"type": "WEB",
"url": "https://gist.github.com/Ro1ME/c85f513d1fd702c686cffe061cbf8ef5"
}
],
"schema_version": "1.4.0",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:H/A:N",
"type": "CVSS_V3"
}
]
}
Mitigation MIT-5.1
Strategy: Input Validation
- Assume all input is malicious. Use an "accept known good" input validation strategy, i.e., use a list of acceptable inputs that strictly conform to specifications. Reject any input that does not strictly conform to specifications, or transform it into something that does.
- When performing input validation, consider all potentially relevant properties, including length, type of input, the full range of acceptable values, missing or extra inputs, syntax, consistency across related fields, and conformance to business rules. As an example of business rule logic, "boat" may be syntactically valid because it only contains alphanumeric characters, but it is not valid if the input is only expected to contain colors such as "red" or "blue."
- Do not rely exclusively on looking for malicious or malformed inputs. This is likely to miss at least one undesirable input, especially if the code's environment changes. This can give attackers enough room to bypass the intended validation. However, denylists can be useful for detecting potential attacks or determining which inputs are so malformed that they should be rejected outright.
- When validating filenames, use stringent allowlists that limit the character set to be used. If feasible, only allow a single "." character in the filename to avoid weaknesses such as CWE-23, and exclude directory separators such as "/" to avoid CWE-36. Use a list of allowable file extensions, which will help to avoid CWE-434.
- Do not rely exclusively on a filtering mechanism that removes potentially dangerous characters. This is equivalent to a denylist, which may be incomplete (CWE-184). For example, filtering "/" is insufficient protection if the filesystem also supports the use of "\" as a directory separator. Another possible error could occur when the filtering is applied in a way that still produces dangerous data (CWE-182). For example, if "../" sequences are removed from the ".../...//" string in a sequential fashion, two instances of "../" would be removed from the original string, but the remaining characters would still form the "../" string.
Mitigation MIT-15
For any security checks that are performed on the client side, ensure that these checks are duplicated on the server side, in order to avoid CWE-602. Attackers can bypass the client-side checks by modifying values after the checks have been performed, or by changing the client to remove the client-side checks entirely. Then, these modified values would be submitted to the server.
Mitigation MIT-20.1
Strategy: Input Validation
- Inputs should be decoded and canonicalized to the application's current internal representation before being validated (CWE-180). Make sure that the application does not decode the same input twice (CWE-174). Such errors could be used to bypass allowlist validation schemes by introducing dangerous inputs after they have been checked.
- Use a built-in path canonicalization function (such as realpath() in C) that produces the canonical version of the pathname, which effectively removes ".." sequences and symbolic links (CWE-23, CWE-59). This includes:
- realpath() in C
- getCanonicalPath() in Java
- GetFullPath() in ASP.NET
- realpath() or abs_path() in Perl
- realpath() in PHP
Mitigation MIT-4
Strategy: Libraries or Frameworks
Use a vetted library or framework that does not allow this weakness to occur or provides constructs that make this weakness easier to avoid [REF-1482].
Mitigation MIT-29
Strategy: Firewall
Use an application firewall that can detect attacks against this weakness. It can be beneficial in cases in which the code cannot be fixed (because it is controlled by a third party), as an emergency prevention measure while more comprehensive software assurance measures are applied, or to provide defense in depth [REF-1481].
Mitigation MIT-17
Strategy: Environment Hardening
Run your code using the lowest privileges that are required to accomplish the necessary tasks [REF-76]. If possible, create isolated accounts with limited privileges that are only used for a single task. That way, a successful attack will not immediately give the attacker access to the rest of the software or its environment. For example, database applications rarely need to run as the database administrator, especially in day-to-day operations.
Mitigation MIT-21.1
Strategy: Enforcement by Conversion
- When the set of acceptable objects, such as filenames or URLs, is limited or known, create a mapping from a set of fixed input values (such as numeric IDs) to the actual filenames or URLs, and reject all other inputs.
- For example, ID 1 could map to "inbox.txt" and ID 2 could map to "profile.txt". Features such as the ESAPI AccessReferenceMap [REF-185] provide this capability.
Mitigation MIT-22
Strategy: Sandbox or Jail
- Run the code in a "jail" or similar sandbox environment that enforces strict boundaries between the process and the operating system. This may effectively restrict which files can be accessed in a particular directory or which commands can be executed by the software.
- OS-level examples include the Unix chroot jail, AppArmor, and SELinux. In general, managed code may provide some protection. For example, java.io.FilePermission in the Java SecurityManager allows the software to specify restrictions on file operations.
- This may not be a feasible solution, and it only limits the impact to the operating system; the rest of the application may still be subject to compromise.
- Be careful to avoid CWE-243 and other weaknesses related to jails.
Mitigation MIT-34
Strategy: Attack Surface Reduction
- Store library, include, and utility files outside of the web document root, if possible. Otherwise, store them in a separate directory and use the web server's access control capabilities to prevent attackers from directly requesting them. One common practice is to define a fixed constant in each calling program, then check for the existence of the constant in the library/include file; if the constant does not exist, then the file was directly requested, and it can exit immediately.
- This significantly reduces the chance of an attacker being able to bypass any protection mechanisms that are in the base program but not in the include files. It will also reduce the attack surface.
Mitigation MIT-39
- Ensure that error messages only contain minimal details that are useful to the intended audience and no one else. The messages need to strike the balance between being too cryptic (which can confuse users) or being too detailed (which may reveal more than intended). The messages should not reveal the methods that were used to determine the error. Attackers can use detailed information to refine or optimize their original attack, thereby increasing their chances of success.
- If errors must be captured in some detail, record them in log messages, but consider what could occur if the log messages can be viewed by attackers. Highly sensitive information such as passwords should never be saved to log files.
- Avoid inconsistent messaging that might accidentally tip off an attacker about internal state, such as whether a user account exists or not.
- In the context of path traversal, error messages which disclose path information can help attackers craft the appropriate attack strings to move through the file system hierarchy.
Mitigation MIT-16
Strategy: Environment Hardening
When using PHP, configure the application so that it does not use register_globals. During implementation, develop the application so that it does not rely on this feature, but be wary of implementing a register_globals emulation that is subject to weaknesses such as CWE-95, CWE-621, and similar issues.
CAPEC-126: Path Traversal
An adversary uses path manipulation methods to exploit insufficient input validation of a target to obtain access to data that should be not be retrievable by ordinary well-formed requests. A typical variety of this attack involves specifying a path to a desired file together with dot-dot-slash characters, resulting in the file access API or function traversing out of the intended directory structure and into the root file system. By replacing or modifying the expected path information the access function or API retrieves the file desired by the attacker. These attacks either involve the attacker providing a complete path to a targeted file or using control characters (e.g. path separators (/ or \) and/or dots (.)) to reach desired directories or files.
CAPEC-64: Using Slashes and URL Encoding Combined to Bypass Validation Logic
This attack targets the encoding of the URL combined with the encoding of the slash characters. An attacker can take advantage of the multiple ways of encoding a URL and abuse the interpretation of the URL. A URL may contain special character that need special syntax handling in order to be interpreted. Special characters are represented using a percentage character followed by two digits representing the octet code of the original character (%HEX-CODE). For instance US-ASCII space character would be represented with %20. This is often referred as escaped ending or percent-encoding. Since the server decodes the URL from the requests, it may restrict the access to some URL paths by validating and filtering out the URL requests it received. An attacker will try to craft an URL with a sequence of special characters which once interpreted by the server will be equivalent to a forbidden URL. It can be difficult to protect against this attack since the URL can contain other format of encoding such as UTF-8 encoding, Unicode-encoding, etc.
CAPEC-76: Manipulating Web Input to File System Calls
An attacker manipulates inputs to the target software which the target software passes to file system calls in the OS. The goal is to gain access to, and perhaps modify, areas of the file system that the target software did not intend to be accessible.
CAPEC-78: Using Escaped Slashes in Alternate Encoding
This attack targets the use of the backslash in alternate encoding. An adversary can provide a backslash as a leading character and causes a parser to believe that the next character is special. This is called an escape. By using that trick, the adversary tries to exploit alternate ways to encode the same character which leads to filter problems and opens avenues to attack.
CAPEC-79: Using Slashes in Alternate Encoding
This attack targets the encoding of the Slash characters. An adversary would try to exploit common filtering problems related to the use of the slashes characters to gain access to resources on the target host. Directory-driven systems, such as file systems and databases, typically use the slash character to indicate traversal between directories or other container components. For murky historical reasons, PCs (and, as a result, Microsoft OSs) choose to use a backslash, whereas the UNIX world typically makes use of the forward slash. The schizophrenic result is that many MS-based systems are required to understand both forms of the slash. This gives the adversary many opportunities to discover and abuse a number of common filtering problems. The goal of this pattern is to discover server software that only applies filters to one version, but not the other.