The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands to orchestrate its research pipeline and run bundled Python validation scripts. It also includes instructions for the agent to publish reports using a project-specific command (pnpm run sync:research) targeting the author's domain (lovstudio.ai), which is a documented vendor-specific functionality.
[INDIRECT_PROMPT_INJECTION]: The skill processes large volumes of untrusted data from the web, which constitutes an indirect prompt injection surface.
Ingestion points: External data is ingested via search tools like search-cli and WebSearch during the retrieval phase.
Boundary markers: The skill methodology (methodology.md) explicitly defines a 'Source trust boundary', instructing the agent to treat web content strictly as data and not as instructions.
Capability inventory: The skill possesses the ability to write files to the disk and execute shell commands for validation and publishing.
Sanitization: The skill employs multiple validation layers, including automated lexical overlap checking in verify_claim_support.py and DOI/URL verification in verify_citations.py to ensure information integrity.
[EXTERNAL_DOWNLOADS]: The documentation recommends installing search-cli from the 199-biotechnologies GitHub repository to enhance search capabilities. This is an optional external tool utilized for its primary research purpose.

deep-research