radar-automation
Pass
Audited by Gen Agent Trust Hub on Mar 29, 2026
Risk Level: SAFE
Full Analysis
- [SAFE]: The skill provides instructions for connecting to established services using official domains (composio.dev, rube.app) and trusted GitHub repositories (ComposioHQ).
- [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface due to its reliance on external tool definitions and execution plans.
- Ingestion points: Dynamic tool schemas and execution plans returned by the RUBE_SEARCH_TOOLS function as described in SKILL.md.
- Boundary markers: No explicit delimiters or instructions to ignore embedded commands are present.
- Capability inventory: The agent can execute a wide range of Radar tools via RUBE_MULTI_EXECUTE_TOOL and RUBE_REMOTE_WORKBENCH.
- Sanitization: The instructions do not describe sanitization or validation logic for external tool schemas.
- [COMMAND_EXECUTION]: The skill implements dynamic tool execution based on runtime search results. This behavior is the primary intended function of the MCP skill and follows standard integration practices.
Audit Metadata