The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or persistence mechanisms were detected. The skill performs standard review tasks within the local environment.
[COMMAND_EXECUTION]: The skill executes standard git diff commands and test runners such as npm test or pytest to analyze implementation against a plan. These actions are within the expected scope of a compliance reviewer.
[PROMPT_INJECTION]: The skill processes project plans and source code, which are ingestion points for potentially untrusted data. 1. Ingestion points: Project plan files (e.g., docs/plans/) and git diff outputs. 2. Boundary markers: Absent; the instructions do not explicitly separate plan content from agent instructions. 3. Capability inventory: Execution of shell commands for git operations and test suites. 4. Sanitization: Absent; the skill relies on the agent's contextual understanding during the review process. This identifies an inherent surface for indirect prompt injection, though no active exploitation was found.

compliance-reviewer