The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted project data to generate documentation, which creates a potential surface for indirect prompt injection. Ingestion points: The agent reads the local project's package.json, tsconfig files, source code, tests, and README. Boundary markers: The skill does not explicitly define boundary markers to isolate file content from the agent's instructions. Capability inventory: The skill uses ripgrep for discovery and is directed to execute existing build, lint, and typecheck scripts found in the repository's package.json. Sanitization: No specific sanitization or filtering of the ingested project data is mentioned.
[COMMAND_EXECUTION]: The skill directs the agent to execute shell commands for project discovery and validation. The agent is instructed to use ripgrep to inventory the public API surface and search for existing documentation tags. Furthermore, the skill directs the agent to run project-specific validation scripts from the local package.json to ensure changes do not break the build, which is a standard developer workflow.

marginalia