opentelemetry-mcp-server

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

• Third-party content exposure detected (high risk: 0.75). Outsider-authored free text can enter the LLM context via runtime retrieval of trace/span fields (e.g., error messages, HTTP URLs/bodies, span attributes) from external backends like Jaeger/Tempo/Traceloop when the MCP tools call get_trace/search_traces and return those readable strings to the agent.

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

• Potentially malicious external URL detected (high risk: 0.90). The MCP is configured to run via "pipx run opentelemetry-mcp", which will fetch and execute the package from PyPI at runtime (https://pypi.org/project/opentelemetry-mcp/), so remote code is retrieved and executed as a required dependency.