gitcode-pr-security-review

Pass

Audited by Gen Agent Trust Hub on Aug 13, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted code content from GitCode Pull Requests, which creates an attack surface for indirect prompt injection where malicious code could attempt to influence the audit results.
  • Ingestion points: Untrusted code patches are retrieved from GitCode PRs in SKILL.md (Step 4.1) and processed via scripts/gitcode_api.py.
  • Boundary markers: Absent. There are no explicit delimiters or instructions to ignore embedded directives when passing PR code to the sdk-security-audit skill or the internal fallback logic.
  • Capability inventory: The skill can create comments on external Pull Requests using the GitCode API (POST /pulls/{number}/comments).
  • Sanitization: No sanitization is performed on the PR code before it is interpreted by the LLM for analysis.
  • Mitigation: The skill implements a human-in-the-loop requirement in SKILL.md (Step 4.5), which requires the user to review and confirm all findings before any comments are posted to the repository.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 13, 2026, 03:24 AM
Security Audit — agent-trust-hub — gitcode-pr-security-review