The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes shell commands to facilitate its research workflow, including calling the search-cli tool for multi-provider web search, executing python scripts for report validation, and running pnpm run sync:research to publish generated reports to a local project directory. These commands are integral to the skill's primary functionality.
[PROMPT_INJECTION]: As the skill ingests and processes large amounts of untrusted data from the web, it possesses an inherent surface for indirect prompt injection. This is addressed in the reference/quality-gates.md file, which establishes a 'Source trust boundary' instructing the agent to treat external content strictly as data and never as instructions, significantly mitigating the risk.
[EXTERNAL_DOWNLOADS]: The skill relies on external utilities for its operation, specifically search-cli and weasyprint. These tools are used for their intended purposes (search and PDF conversion) and are clearly identified in the installation documentation.

deep-research