The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The pr-prepare agent processes untrusted data from the user's local repository, including source code, documentation, and commit history, to perform its evaluation. This ingestion of external content constitutes an indirect prompt injection surface.
Ingestion points: The agent reads project files such as CONTRIBUTING.md, package.json, and all source code files targeted by Grep or Read operations in agents/pr-prepare.md.
Boundary markers: No specific delimiters or instructions to ignore embedded commands are used to encapsulate the data retrieved from the repository.
Capability inventory: The agent possesses the Bash, Read, Grep, and Glob tools, and it is designed to execute project-defined scripts such as npm test and npm run lint.
Sanitization: The skill does not implement explicit sanitization or filtering of file contents before they are processed by the agent's logic.

open-source-contributions