The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted local data from the repository state to generate pull request metadata, which could theoretically be manipulated by malicious file content or commit messages. * Ingestion points: git diff, git log, and git status calls referenced in SKILL.md. * Boundary markers: Not present for the ingested repository content. * Capability inventory: git push, gh pr create, and gh api calls in SKILL.md. * Sanitization: Relies on agent validation and mandatory user confirmation rather than automated sanitization.
[SAFE]: No malicious patterns or security vulnerabilities were identified. The instructions prioritize the protection of sensitive information, such as credentials and customer data, and enforce human-in-the-loop authorization for all external modifications.

create-pr