The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external sources (GitHub issues and pull request comments) which could potentially contain malicious instructions.
Ingestion points: Uses gh issue view <N> --comments and gh pr view <N> --comments to fetch external content into the agent's context (SKILL.md).
Boundary markers: No specific delimiters or instructions are provided to the agent to treat the fetched content as data rather than instructions.
Capability inventory: The skill possesses extensive capabilities including shell command execution (git, gh, lsof, pg0, kill), file system access, and branch management.
Sanitization: There is no evidence of sanitization or filtering of the fetched GitHub content before it is analyzed by the agent.
[COMMAND_EXECUTION]: The skill instructions involve executing a wide range of shell commands to facilitate debugging and reproduction of issues.
Evidence: Commands such as git log, git blame, gh issue list, lsof, and pg0 are used to investigate the repository state and manage local development services.
Execution of local scripts: The skill explicitly directs the agent to run benchmark scripts (e.g., ./scripts/benchmarks/, hindsight-dev/benchmarks/), which is a standard procedure for performance triage but involves executing code from the repository.

triage