The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to autonomously research codebases, which involves ingesting potentially untrusted data from source files, manifests, and existing documentation.
Ingestion points: The project research phase (Phase 1) reads package manifests, directory trees, entry points, source files, tests, and configuration files (including .env and .config.js) into the agent context.
Boundary markers: The instructions do not specify the use of delimiters or specific "ignore" instructions when the agent processes untrusted codebase content to prevent accidental execution of embedded instructions.
Capability inventory: The skill has broad capabilities including file system writes (creating/updating docs), shell script execution (scripts/find-docs.sh), package management (npm install), and cloud deployment (vercel --prod).
Sanitization: While continuous-docs.md instructs the agent not to record secrets as facts, there is no explicit sanitization or filtering of ingested data before it influences the agent's output or actions.
[EXTERNAL_DOWNLOADS]: The skill references several external utility skills (such as unslop, stop-slop, and use-afs) and provides commands to install them via a local Python script (install-external-skills.py). These resources trace back to the same author (alvarovillalbaa) and are part of the intended project workflow.
[COMMAND_EXECUTION]: The skill utilizes a helper script (scripts/find-docs.sh) to discover documentation paths and provides instructions for executing build and deployment tools (npm, vercel) as part of the documentation site generation workflow.

code-documentation