mcp-exa-search

Fail

Audited by Gen Agent Trust Hub on Mar 13, 2026

Risk Level: HIGHREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [REMOTE_CODE_EXECUTION]: The installation instructions include a command to pipe a remote shell script from the author's GitHub repository directly to the system shell (curl -fsSL ... | sh). While this is a resource provided by the author to support the skill's functionality, this method of installation bypasses standard package manager safety checks.
  • [COMMAND_EXECUTION]: The skill uses the mh command-line utility to interact with the Exa AI service and perform its intended tasks.
  • [PROMPT_INJECTION]: The skill processes external web content, which presents a surface for indirect prompt injection. 1. Ingestion points: Results from webSearchExa and getCodeContextExa (SKILL.md). 2. Boundary markers: None identified in the provided instructions. 3. Capability inventory: Local command execution via the mh CLI. 4. Sanitization: No specific sanitization or filtering of search results is mentioned.
Recommendations
  • HIGH: Downloads and executes remote code from: https://raw.githubusercontent.com/vaayne/mcphub/main/scripts/install.sh - DO NOT USE without thorough review
Audit Metadata
Risk Level
HIGH
Analyzed
Mar 13, 2026, 02:02 PM