The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses npx, pnpm exec, yarn exec, or bunx to download and execute the agent-reviews package from the NPM registry. This package is a vendor-owned resource authored by pbakaus.
[COMMAND_EXECUTION]: The skill invokes several high-privilege commands including git push to upload code changes to a remote repository and git config --global to modify user identity in cloud environments. It also executes the agent-reviews CLI for replying to and resolving PR threads.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface (Category 8). It is designed to ingest human PR comments and 'Act on Evaluation' by fixing code based on that input.
Ingestion points: Human review comments fetched from GitHub via the agent-reviews --expanded command.
Boundary markers: Absent. There are no delimiters or specific instructions provided to the agent to treat comment text as untrusted or to ignore embedded instructions.
Capability inventory: The agent can modify the local file system, perform git commit, and execute git push to the remote repository.
Sanitization: No validation or sanitization of the comment body is performed before the agent attempts to interpret and implement the suggested changes.

resolve-human-reviews