The Agent Skills Directory

[SAFE]: The skill configuration and tool definitions do not exhibit malicious patterns. Setup instructions for the Exa MCP server use standard placeholders for sensitive credentials and reference well-known, legitimate search services.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests untrusted data from the public web. Ingestion points: web_search_exa and get_code_context_exa in SKILL.md. Boundary markers: Absent. Capability inventory: Neural web search and code context retrieval. Sanitization: Absent. This surface is inherent to the primary function of a search-based skill and is managed by the AI agent's internal content safety measures.

exa-search