review
Pass
Audited by Gen Agent Trust Hub on Oct 11, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data, including code diffs and requirement specifications, which serves as a potential vector for indirect prompt injection attacks.\n
- Ingestion points: The skill reads commit logs, git diffs, and specification documents from various files including
spec.md,.agent/context/issue-tracker.md, and project documentation repositories.\n - Boundary markers: The instructions attempt to mitigate risk by requiring the agent to use a structured output format (Conventional Comments) and providing specific evidence-based criteria for its findings.\n
- Capability inventory: The skill has the ability to execute git commands and write to local documentation files (the "knowledge vault").\n
- Sanitization: There is no explicit instruction for the agent to sanitize or escape the content of diffs or specifications before they are processed in the prompt context.\n- [COMMAND_EXECUTION]: The skill utilizes shell-based commands to interact with the repository's version control system.\n
- Evidence: In
SKILL.md, the instructions require the execution ofgit rev-parse,git diff, andgit logto extract repository state and history for analysis.
Audit Metadata