The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute various Git commands including git status, git diff, git add, git commit, git push, and git pull. These are necessary for the skill's stated purpose of automating version control workflows.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it ingests untrusted data from the repository, specifically via git diff and git log, which are then used to draft commit messages.
Ingestion points: Reads repository history and current changes in SKILL.md (via git diff and git log).
Boundary markers: The instructions specify analyzing the diff to draft a message, but do not define strict delimiters for the untrusted content.
Capability inventory: The skill can execute subprocesses (git commands) and perform network operations (git push).
Sanitization: The skill includes a specific rule to skip sensitive files such as .env and credentials* during the git add phase, reducing the risk of accidental data leakage.
[DATA_EXFILTRATION]: The skill performs network operations via git push to upload local code to a remote repository. This is the intended primary purpose of the skill and includes a safety check to prevent accidental pushes to protected branches (main, master, dev, test) by requiring user confirmation via platform-specific interactive tools.

cp