The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands using the GitHub CLI (gh) and git to manage pull request workflows. It also dynamically invokes project-specific testing and linting tools (e.g., npm test, make) based on the repository's configuration files discovered during the checkout process.
[PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks because it processes untrusted data from pull request titles, descriptions, and file contents.
Ingestion points: PR metadata, commit messages, and diffs are fetched in scripts/fetch-pr.sh and processed throughout the workflow in SKILL.md.
Boundary markers: The skill does not use specific delimiters or instructions to isolate the untrusted PR content from the agent's internal logic.
Capability inventory: The skill has the ability to execute shell commands, read/write files in the local repository, and post comments back to GitHub via the gh API.
Sanitization: There is no evidence of sanitization or filtering of the content retrieved from the pull request before it is analyzed by the agent.

review-pr