The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes standard Git commands including git diff, git log, and git rev-parse to manage code comparison. These commands are restricted to the local filesystem and intended repository metadata.
[DATA_EXFILTRATION]: No exfiltration vectors are present. The skill does not communicate with external servers or access sensitive credential files such as .env or .ssh keys.
[PROMPT_INJECTION]: The skill implements an indirect prompt injection surface by ingesting untrusted data from git diff outputs and specification files. Evidence chain: (1) Ingestion points: SKILL.md reads Git diffs and specs/ files; (2) Boundary markers: Absent; (3) Capability inventory: git commands and subagent spawning; (4) Sanitization: Absent. The risk is classified as low as this behavior is core to the functionality of a code review tool.

pb-requesting-code-review