Loading the catalogue…
Agent frameworks · LangChain Inc. · US
LangChain
US CLOUD Act exposureOpen sourceMIT
STAV REVIEW · LAST REVIEWED 06/21/2026
“The dominant open-source agent framework for Sovereign inference orchestration, but six CVEs in 90 days — one still unresolved — demand active security discipline.”
Good for
- Teams building multi-step inference pipelines who want full control over which providers handle their data — LangChain Inc. is entirely outside the default data path
- Python and TypeScript shops who need a battle-tested, MIT-licensed framework with 320M monthly PyPI downloads and genuine production entrenchment
- Organisations that require self-hosted or on-premises observability, as LangSmith supports both deployment modes alongside SSO and audit logging
- Teams on Stav who want to wire Smart Router–managed endpoints into composable chains and agents without a mandatory vendor proxy
Watch out for
- An unresolved path-traversal and sandbox-escape advisory (CVE-2026-55443, published 2026-06-12) makes new onboarding in regulated environments inadvisable until a confirmed patch ships — this is the second structurally similar path-traversal flaw in the same 90-day window, signalling architectural risk in file-handling and deserialization subsystems that isolated releases have not yet closed
- No public bug bounty, no security.txt, no SBOM, and unsigned PyPI packages — responsible disclosure flows only through GitHub Security Advisories, so teams must subscribe to that feed and enforce a rapid patch-update policy
- Enabling LangSmith tracing (LANGCHAIN_TRACING_V2) routes prompts and completions to a US-operated endpoint subject to CLOUD Act compelled disclosure; trace-retention periods, SOC 2 status, and DPA availability are all publicly unconfirmed and must be resolved with LangChain Inc. before production use
Editorial assessment, not legal advice. Stav's risk ratings, scores, and verdicts are our own analysis of publicly available information and may be incomplete or out of date. Verify independently before making compliance or procurement decisions.
MODEL REQUIREMENTS
What LangChain needs from a model.
Required
None catalogued.
Recommended
- JSON mode
- Long context (≥128k)
- Structured output
- Tool calling
Known issues8 open
Public security, privacy and governance advisories that are still open. Each row carries a source link — these are factual claims, not editorial judgements.
2026-06-30
lowUpstream change on langchain: github_last_push '2026-06-23T04:34:02Z' → '2026-06-29T19:47:27Z'
other · open
Source ↗2026-06-23
lowUpstream change on langchain: github_release 'langchain-core==1.4.7' → 'langchain-core==1.4.8', github_last_push '2026-06-16T00:12:47Z' → '2026-06-23T04:34:02Z'
other · open
Source ↗2026-06-16
lowUpstream change on langchain: github_release 'langchain-core==1.4.2' → 'langchain-core==1.4.7', github_last_push '2026-06-08T18:19:52Z' → '2026-06-16T00:12:47Z', cve_count 8 → 9
other · open
Source ↗2026-06-09
lowUpstream change on langchain: github_release 'langchain-core==1.4.0' → 'langchain-core==1.4.2', github_last_push '2026-06-02T00:04:01Z' → '2026-06-08T18:19:52Z'
other · open
Source ↗2026-06-02
lowUpstream change on langchain: github_last_push '2026-05-19T03:23:28Z' → '2026-06-02T00:04:01Z'