Skills
Navigate
skills/jeremylongshore/tons-of-skills-marketplace/langchain-ci-integration/Gen Agent Trust Hub

langchain-ci-integration

Pass

Audited by Gen Agent Trust Hub on Sep 9, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The scripts/run_eval.py script demonstrates an indirect prompt injection surface by ingesting JSON results from an evaluation harness and posting them as GitHub PR comments.
  • Ingestion points: JSON output from eval_harness in scripts/run_eval.py.
  • Boundary markers: No explicit delimiters are used when formatting the markdown comment.
  • Capability inventory: Write capability used to post pull request comments via the GitHub CLI.
  • Sanitization: Data is directly interpolated into markdown; however, the risk is limited to the reporting context.
  • [DYNAMIC_EXECUTION]: The skill uses dynamic module loading in scripts/dryrun_load_chains.py to verify chain compatibility.
  • Implementation: Uses importlib.import_module to load project files.
  • Scope: Restricts loading to the local src/chains directory, serving as a functional test for import errors.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 9, 2026, 03:45 AM