exa-web-search-free

Pass

Audited by Gen Agent Trust Hub on Mar 29, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill connects to a remote Model Context Protocol (MCP) server at https://mcp.exa.ai/mcp. This is the official endpoint provided by Exa, a well-known AI search service.
  • [COMMAND_EXECUTION]: Utilizes the mcporter CLI tool to perform searches and configure the MCP server connection.
  • [PROMPT_INJECTION]: Indirect prompt injection surface identified through web data ingestion.
  • Ingestion points: Untrusted content enters the agent context from web search results, code documentation snippets, and crawled web pages.
  • Boundary markers: The instructions do not specify explicit markers or instructions to isolate the retrieved search content from the agent's core instructions.
  • Capability inventory: The skill performs network-based information retrieval; the agent's response to ingested data depends on its available toolset (e.g., file system access or code execution).
  • Sanitization: The skill does not implement filtering or sanitization of the content retrieved from the internet.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 29, 2026, 02:05 AM
Security Audit — agent-trust-hub — exa-web-search-free