The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent to locate and execute arbitrary scripts found within the target repository, including those in the scripts/ directory, Makefile targets, and manifest script blocks (e.g., package.json). This grants the agent the ability to run code defined by the repository's contents as part of the lockfile regeneration and verification process.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection due to the processing of untrusted data from external pull requests.
Ingestion points: Data is ingested via gh pr list and gh pr diff in the 'Read the incoming PRs' section of SKILL.md to collect package versions and diff content.
Boundary markers: The instructions do not define delimiters or provide warnings to the agent to ignore instructions embedded within the pull request data.
Capability inventory: The agent possesses shell execution capabilities for git, gh, package managers (npm, pnpm, yarn, bun), and repository-specific build/test tools.
Sanitization: There is no mention of sanitizing or validating pull request titles, descriptions, or code diffs before processing them.

batch-dependency-bumps