The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. It is instructed to extract migration guidelines (API changes, affected files, and required fixes) from external GitHub PR comments that match a specific header ('## Dependency Upgrade Review'). A malicious actor could post a comment with this header to influence the agent's actions or induce it to execute unintended commands.
Ingestion points: PR comments fetched via gh pr view in SKILL.md.
Boundary markers: None; the skill uses simple string matching to identify instructions and does not wrap external data in delimiters to prevent instruction override.
Capability inventory: The skill has access to the Bash tool and performs git commit, git push, gh pr comment, and runs local pnpm scripts.
Sanitization: The skill does not sanitize or validate the content of the PR comments before treating them as authoritative instructions for code modifications.
[COMMAND_EXECUTION]: The skill executes shell commands and project-defined scripts (pnpm run build, pnpm run format). While standard for development workflows, this involves executing arbitrary code defined in the repository's configuration files.
[EXTERNAL_DOWNLOADS]: The skill uses the GitHub CLI (gh) to fetch data from remote PR comments. This is an expected operation involving a well-known service for the skill's stated purpose.

renovate-migrate