The Agent Skills Directory

[PROMPT_INJECTION]: The skill processes untrusted user data, including commit messages, PR titles, and PR descriptions, which presents a surface for indirect prompt injection.
Ingestion points: The agent reads user-provided text through git log, git show, and PR metadata as described in SKILL.md.
Boundary markers: The validation logic does not explicitly define delimiters to separate untrusted content from the agent's instructions.
Capability inventory: The agent has the ability to execute shell commands, perform Git operations, and use the GitHub CLI (gh).
Sanitization: No sanitization or escaping of external strings is mentioned before the agent evaluates them.
[EXTERNAL_DOWNLOADS]: The skill recommends the installation of standard development dependencies from the npm registry.
Recommends @commitlint/cli, @commitlint/config-conventional, and husky for repository enforcement.
References the use of the amannn/action-semantic-pull-request GitHub Action in CI workflows.
[COMMAND_EXECUTION]: The skill involves the generation and execution of shell scripts and commands for repository management.
The setup process involves writing shell commands to Git hook files (e.g., .husky/commit-msg).
The implementation notes describe using standard CLI tools like grep, sed, and git to audit the repository state.
Provides instructions to run a local setup script (./setup.sh) to activate repository hooks.

the-doorman