The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill identifies and processes external data (code diffs) which may contain malicious instructions designed to influence the agent's behavior during the commit process.
Ingestion points: SKILL.md (the skill uses git diff and git diff --cached to ingest code changes into the agent's context).
Boundary markers: Absent (there are no instructions to the agent to disregard instructions found within the code changes being reviewed).
Capability inventory: SKILL.md (the agent is authorized to execute git add -A and is instructed to run shell-based tools for linting, type-checking, and testing to resolve errors).
Sanitization: Absent (the skill does not define filters or sanitization for the content of the diffs).

git-commit