The Agent Skills Directory

[COMMAND_EXECUTION]: The skill dynamically identifies and executes shell commands for linting, formatting, and type-checking by reading values from the repository's package.json scripts and external documentation files (AGENTS.md, CLAUDE.md). This creates a dynamic execution path based on external project data.
[COMMAND_EXECUTION]: The skill performs automated Git operations, including staging all changes and committing them with generic messages, which could lead to unintended modifications of the source code.
[PROMPT_INJECTION]: The skill is vulnerable to Indirect Prompt Injection (Category 8c: Tool output poisoning).
Ingestion points: Reads commands and conventions from AGENTS.md, CLAUDE.md, and package.json in the current working directory.
Boundary markers: Absent. The skill instructions do not specify any delimiters or safety checks for the content read from these files.
Capability inventory: Access to the Bash tool allowing execution of package managers (pnpm, npm, yarn), the GitHub CLI (gh), and generic shell utilities (grep, git).
Sanitization: None detected. The skill directly executes commands extracted from the configuration files.
[DATA_EXFILTRATION]: Uses the GitHub CLI (gh pr list) to retrieve information from the GitHub service. This targets a well-known service and is consistent with the skill's stated purpose of reviewing Renovate PRs.

repo-maintenance-node