The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest untrusted data from external sources (documentation sites, GitHub repositories, PDFs, videos) and process it into AI-ready skills. This presents a vulnerability surface where instructions embedded in the source material could potentially influence the agent's behavior during the conversion or enhancement phase.
Ingestion points: scrape_docs, scrape_github, scrape_pdf, scrape_video, scrape_codebase, and scrape_generic tools ingest data from arbitrary external and local sources in SKILL.md.
Boundary markers: No explicit delimiters or 'ignore embedded instructions' warnings are described in the skill workflow.
Capability inventory: The skill uses tools for scraping (network/file read), post-processing (enhance_skill), and installation (install_skill, upload_skill).
Sanitization: The skill description does not specify how it sanitizes or filters potentially malicious instructions from the ingested content.
[EXTERNAL_DOWNLOADS]: The skill facilitates downloading content from arbitrary documentation websites, GitHub repositories, and video platforms via tools like scrape_docs, scrape_github, and scrape_video.
[COMMAND_EXECUTION]: The skill includes an install_skill tool and a scrape_codebase tool, which imply interaction with the local system environment and potentially the execution of installation scripts or code analysis tools.

skill-builder