The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses standard git and bash commands for developer workflows. It includes a 'Git Safety Protocol' that prohibits destructive actions like force pushing to main or updating git configuration without request.
[DATA_EXPOSURE]: The skill explicitly instructs the agent to never commit sensitive files such as .env, credentials.json, or private keys, following security best practices.
[PROMPT_INJECTION]: An indirect prompt injection surface is present as the skill ingests untrusted data via git diff and git status to generate commit messages. Malicious content within the codebase could theoretically influence the agent's message generation. Evidence: 1. Ingestion points: git diff and git status in SKILL.md. 2. Boundary markers: Absent. 3. Capability inventory: git add and git commit in SKILL.md. 4. Sanitization: Absent.

git-commit