brew-langgraph-cli-cve-2026-59152
Vulnerability from osv_homebrew
Summary
An attacker who can send an HTTP request to a server running the LangSmith SDK's TracingMiddleware can cause that server to read an arbitrary file from its local filesystem and upload the contents to LangSmith as a trace attachment. Depending on how the distributed trace system is deployed, triggering a read may not require authentication. Retrieving the contents requires read access to the LangSmith workspace the traces are sent to. The net effect is a trust-boundary crossing: a party with workspace trace-read access (for example a low-privilege workspace member, a contractor, or a compromised teammate account) gains the ability to read files from any server running TracingMiddleware, a capability outside that workspace's intended trust boundary.
Impact
Confidentiality (High): arbitrary read of files accessible to the server process, exposed to anyone with workspace trace-read access.
Details
Two defects combine. A field supplied through a tracing-propagation header was merged into the run without validation, allowing injection of run attributes including attachments (CWE-346). A type check intended to gate filesystem access did not match the type of the decoded input, so the guard never engaged (CWE-843). As a result, an attacker-named file is opened by the server and uploaded as a trace attachment by the background tracing thread (CWE-22).
Who can exploit this
- Anyone reachable by HTTP can trigger the file read. Depending on how the distributed trace system is deployed, triggering may not require authentication.
- Retrieving the file contents requires read access to the destination LangSmith workspace. The upload uses the server's own configured API key and workspace, which the attacker cannot redirect, so a zero-access outsider cannot retrieve the result; a workspace member, or anyone who has compromised one, can.
Remediation
Upgrade the Python SDK to >= 0.8.18.
Workarounds
Until upgrading, do not expose TracingMiddleware to untrusted HTTP traffic, and limit workspace trace-read access to trusted members.
Credits
First reported by @Ryu7zz.
{
"affected": [
{
"ecosystem_specific": {
"fix": "bump",
"range_state": "fixed",
"resource": "langsmith",
"resource_purl": "pkg:pypi/langsmith@0.10.2",
"upstream_fixed_in": "0.8.18"
},
"package": {
"ecosystem": "Homebrew",
"name": "langgraph-cli",
"purl": "pkg:brew/langgraph-cli"
},
"ranges": [
{
"events": [
{
"introduced": "0"
},
{
"fixed": "0.4.31"
}
],
"type": "ECOSYSTEM"
}
]
}
],
"database_specific": {
"confidence": "high",
"source": "matched",
"strategy": "registry",
"upstream_evidence": [
{
"ecosystem": "PyPI",
"key": "pkg:pypi/langsmith@0.10.2",
"name": "langsmith",
"resource": "langsmith",
"strategy": "registry",
"subject_version": "0.10.2"
}
]
},
"details": "# Summary\n\nAn attacker who can send an HTTP request to a server running the LangSmith SDK\u0027s `TracingMiddleware` can cause that server to read an arbitrary file from its local filesystem and upload the contents to LangSmith as a trace attachment. Depending on how the distributed trace system is deployed, triggering a read may not require authentication. Retrieving the contents requires read access to the LangSmith workspace the traces are sent to. The net effect is a trust-boundary crossing: a party with workspace trace-read access (for example a low-privilege workspace member, a contractor, or a compromised teammate account) gains the ability to read files from any server running `TracingMiddleware`, a capability outside that workspace\u0027s intended trust boundary.\n\n# Impact\n\nConfidentiality (High): arbitrary read of files accessible to the server process, exposed to anyone with workspace trace-read access.\n\n# Details\n\nTwo defects combine. A field supplied through a tracing-propagation header was merged into the run without validation, allowing injection of run attributes including attachments (CWE-346). A type check intended to gate filesystem access did not match the type of the decoded input, so the guard never engaged (CWE-843). As a result, an attacker-named file is opened by the server and uploaded as a trace attachment by the background tracing thread (CWE-22).\n\n## Who can exploit this\n\n- Anyone reachable by HTTP can trigger the file read. Depending on how the distributed trace system is deployed, triggering may not require authentication.\n- Retrieving the file contents requires read access to the destination LangSmith workspace. The upload uses the server\u0027s own configured API key and workspace, which the attacker cannot redirect, so a zero-access outsider cannot retrieve the result; a workspace member, or anyone who has compromised one, can.\n\n# Remediation\n\nUpgrade the Python SDK to `\u003e= 0.8.18`.\n\n# Workarounds\n\nUntil upgrading, do not expose `TracingMiddleware` to untrusted HTTP traffic, and limit workspace trace-read access to trusted members.\n\n# Credits\n\nFirst reported by @Ryu7zz.",
"id": "BREW-langgraph-cli-CVE-2026-59152",
"modified": "2026-08-13T17:02:14Z",
"published": "2026-08-13T17:02:14Z",
"references": [
{
"type": "WEB",
"url": "https://github.com/langchain-ai/langsmith-sdk/security/advisories/GHSA-f4xh-w4cj-qxq8"
},
{
"type": "PACKAGE",
"url": "https://github.com/langchain-ai/langsmith-sdk"
}
],
"schema_version": "1.7.3",
"severity": [
{
"score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:H/I:N/A:N",
"type": "CVSS_V3"
}
],
"summary": "LangSmith SDK TracingMiddleware: Arbitrary server-side file read",
"upstream": [
"GHSA-f4xh-w4cj-qxq8",
"CVE-2026-59152"
]
}
Sightings
| Author | Source | Type | Date | Other |
|---|
Nomenclature
- Seen: The vulnerability was mentioned, discussed, or observed by the user.
- Confirmed: The vulnerability has been validated from an analyst's perspective.
- Published Proof of Concept: A public proof of concept is available for this vulnerability.
- Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
- Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
- Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
- Not confirmed: The user expressed doubt about the validity of the vulnerability.
- Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.
Related by attack behaviour
Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.