The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local developer commands including git and pnpm (pnpm lint, pnpm build, git commit) to analyze and update the repository. These operations are essential for the skill's stated purpose of automating commits.
[DATA_EXPOSURE]: The skill reads the repository's staged changes via git diff. It contains a specific safety rule to identify and warn the user before staging potentially sensitive items such as .env files, credentials, or secrets, which demonstrates a proactive security posture.
[PROMPT_INJECTION]: The skill includes an indirect prompt injection surface (Category 8) as it processes untrusted code changes from the repository. However, it mitigates risk by using a HEREDOC (EOF) pattern when executing the git commit command, ensuring that the generated message is treated as a string and not as executable shell code.
Ingestion points: Reads repository data via git diff --staged (SKILL.md).
Boundary markers: Utilizes HEREDOC (EOF) for shell command isolation.
Capability inventory: Executes local subprocesses for pnpm and git (SKILL.md).
Sanitization: Relies on the user to approve changes and uses secure string literal wrapping in the CLI.

commit