mcp-servers

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill explicitly fetches and ingests external OpenAPI specs and external agent card URLs at runtime (see "Register MCP Server (OpenAPI)" in SKILL.md and the manifest/schema notes that "Remote OpenAPI specs are fetched at runtime" and that agent_card_url/hosted-a2a-agent sources are fetched), which are untrusted third‑party resources that can be converted into tools or influence agent behavior.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 1.00). The skill explicitly fetches remote OpenAPI specs at runtime (e.g., https://internal-api.example.com/openapi.json and https://api.weather.example.com/openapi.json) which are auto-converted into MCP tools that directly control agent capabilities (and it also documents runtime-fetching of external agent cards like https://research-agent.example.com/.well-known/agent.json), so these runtime-fetched URLs can directly influence agent prompts/behavior.