add-mcp-from-remote-url

Pass

Audited by Gen Agent Trust Hub on Aug 27, 2026

Risk Level: SAFE
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from user-supplied remote URLs, which represents a potential attack surface. However, it incorporates several layers of mitigation:
  • Ingestion points: Data from the remote_url is ingested via the inspect_mcp_candidate tool as described in the workflow (SKILL.md).
  • Boundary markers: The instructions require the agent to present tool counts, names, and authentication posture to the user for explicit confirmation before proceeding.
  • Capability inventory: The skill allows for MCP registration and identity provider attachment, but these are gated by user-mediated steps.
  • Sanitization: The workflow relies on the inspect_mcp_candidate tool to return read-only 'bounded evidence' to prevent immediate execution of untrusted content.
  • [DATA_EXPOSURE]: The skill establishes strict guidelines to prevent credential harvesting, explicitly instructing the agent never to ask for API keys, tokens, or client secrets within the chat environment. All sensitive authentication is offloaded to a secure external dashboard.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 27, 2026, 04:04 PM
Security Audit — agent-trust-hub — add-mcp-from-remote-url