lesson-learned

Pass

Audited by Gen Agent Trust Hub on Sep 16, 2026

Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill instructs the agent to run several git commands, including git log, git diff, and git show, to retrieve commit history and code changes. It uses user-provided inputs such as commit SHAs or ranges to scope these commands, which necessitates proper sanitization by the underlying platform to prevent shell injection.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests and analyzes untrusted data from the repository's git logs and diffs to identify engineering principles. This creates a surface where malicious instructions hidden in commit messages or code comments could influence the agent's output.
  • Ingestion points: Git commit messages and diff content are retrieved in SKILL.md (Phase 2).
  • Boundary markers: No specific delimiters or instructions to ignore embedded commands are defined for the ingested data.
  • Capability inventory: The skill has the capability to read repository history and file changes.
  • Sanitization: The skill does not perform sanitization on the content of the diffs or commit messages before analysis.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 16, 2026, 04:43 PM
Security Audit — agent-trust-hub — lesson-learned