The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local Git repository, specifically the output of git diff --cached and git log, to draft commit messages. An attacker with the ability to commit code or comments to the repository could attempt to embed instructions designed to manipulate the agent's output or actions.
Ingestion points: git diff --cached and git log --oneline -20.
Boundary markers: No explicit boundary markers or delimiters are defined in the instructions to isolate the ingested Git data from the agent's operational instructions.
Capability inventory: Shell command execution (git, npm, chmod, npx) and local file creation (commitlint.config.js).
Sanitization: The skill provides instructions to avoid inventing changes based on dialogue and to use only actual Git diffs as evidence for commit content.
[COMMAND_EXECUTION]: The skill facilitates the execution of shell commands for environment setup and version control management, including installing development dependencies via npm from well-known registries, initializing husky, and setting execution permissions on Git hooks using chmod +x.

codeguard-git-commit