The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external code changes, PR descriptions, and project-specific principle files which could contain hidden instructions designed to influence the agent.
Ingestion points: Reads brain/principles.md, follows linked principle files, and gathers code/plans under review in Step 3.
Boundary markers: The instructions do not define specific delimiters or warnings to separate external data from the skill's operational logic.
Capability inventory: The skill has the ability to spawn subagents (Explore), execute git commands, and invoke other local domain skills.
Sanitization: There is no evidence of filtering or sanitization of the content from files under review.
[DYNAMIC_EXECUTION]: The skill discovers and invokes other installed skills dynamically based on the code domain being analyzed.
Evidence: Step 4 explicitly directs the agent to check .agents/skills/ and .claude/skills/ and "Invoke matched skills now".
[COMMAND_EXECUTION]: The skill uses shell commands to extract repository context and history.
Evidence: Step 5 executes git diff --stat and git log --oneline to validate the scope of the review.

review