The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted content from git log (commit messages) and git diff (source code changes) to perform its analysis. This content could be crafted by an attacker to include hidden instructions designed to influence the agent's summary or verdict.
Ingestion points: Git commit messages and code diffs are fetched and read in SKILL.md (Step 2).
Boundary markers: Absent. The agent is directed to read and internalize the diff and log content without delimiters or specific instructions to ignore embedded commands.
Capability inventory: The skill utilizes bash, git, and basic file operations (read, write, edit).
Sanitization: No validation or filtering is performed on the data fetched from the git repository before it is processed by the agent.
[COMMAND_EXECUTION]: The skill uses shell commands to interact with the local repository and the remote origin.
Evidence: SKILL.md contains several bash blocks executing git remote, git fetch, git log, git diff, and git branch -D to manage the PR review workflow.

pr-grill-me