The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill uses uv run --with exa-py to execute its scripts, which involves downloading the official exa-py package from a standard registry. This is a secure method for managing dependencies for a well-known service.
[SAFE]: Authentication is handled correctly via the EXA_API_KEY environment variable. The skill includes guidance on using .env files for local secret management, which is a recommended practice.
[SAFE]: Network activity is limited to the legitimate Exa API endpoints for search and extraction, fulfilling the skill's stated purpose without suspicious data exfiltration.
[SAFE]: While the skill ingests content from external web pages, it provides the agent with structured instructions to parse the output and synthesize responses with citations, reducing risks associated with untrusted data.
[SAFE]: The instruction templates use standard shell command patterns with placeholders for user arguments. There are no patterns suggesting malicious command injection or dynamic execution from untrusted sources.

exa-search