The Agent Skills Directory

[SAFE]: The skill implements a secure git commit workflow that emphasizes manual review and granular staging.
[SAFE]: It incorporates explicit security sanity checks, instructing the agent to verify the absence of secrets, API tokens, or accidental debug logs before finalizing a commit.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository data through git diff and executes local verification commands like tests or build scripts.
Ingestion points: SKILL.md (Reviewing code changes via git diff and git diff --cached).
Boundary markers: Not explicitly defined for the diff content.
Capability inventory: Execution of git commands and local repository verification scripts (tests/lint/build).
Sanitization: None.
Note: This surface is inherent to the skill's primary function and is mitigated by the structured review process.
[COMMAND_EXECUTION]: Utilizes standard git tooling and local scripts for code management. No malicious or unauthorized command patterns were identified.

commit-work