mcp-setup

Pass

Audited by Gen Agent Trust Hub on Feb 17, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • EXTERNAL_DOWNLOADS (LOW): The skill automates the installation of MCP servers using npx and docker. It references packages like @upstash/context7-mcp and exa-mcp-server. While these are common in the MCP ecosystem, they involve executing code from remote registries.
  • COMMAND_EXECUTION (LOW): The skill generates and executes shell commands using the claude mcp add CLI. The 'Custom' configuration step allows the agent to execute arbitrary commands and arguments provided during the interaction, which could be exploited if malicious inputs are processed.
  • CREDENTIALS_UNSAFE (SAFE): The skill correctly identifies the need for API keys (Exa, GitHub) and suggests passing them via environment variables (-e) to the CLI. It does not hardcode or exfiltrate these secrets.
  • INDIRECT_PROMPT_INJECTION (LOW): The skill defines a clear ingestion surface for user-provided configuration (paths, keys, and custom commands). While it lacks explicit sanitization markers, the risk is mitigated by the fact that the agent is performing an intended configuration task.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 17, 2026, 05:21 PM