The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external community registries and aggregators (e.g., clawhub.ai, skillsmp.com). Maliciously crafted metadata or descriptions in third-party skills could potentially influence the agent's ranking or interpretation of results. * Ingestion points: WebFetch operations performed on external domains (skills.sh, clawhub.ai, skillsmp.com) to retrieve skill listings. * Boundary markers: No explicit delimiters are specified for the retrieved external text in the search logic. * Capability inventory: No dangerous system-level capabilities detected; the skill is limited to searching and formatting recommendations. * Sanitization: The skill proactively instructs the agent to perform security checks on search results, filtering out skills that request sensitive permissions or contain dynamic code execution patterns.

search-skill