mcp-query-router
Pass
Audited by Gen Agent Trust Hub on Mar 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill uses the
rwCLI to perform operations such asrw company,rw chain, andrw macro. It also supports a genericrw callcommand for direct tool invocation. - [DATA_EXPOSURE_AND_EXFILTRATION]: The skill references a specific local file path (
~/Max/research-warehouse/docs/query-guide.md) for validating its parameter schema. - [INDIRECT_PROMPT_INJECTION]: The skill processes user research requests to route queries, creating a surface where malicious input could influence tool parameters.
- Ingestion points: User requests are mapped to tool domains and parameters in
SKILL.md. - Boundary markers: No explicit delimiters or guardrail instructions are provided to isolate user-supplied strings within the CLI arguments.
- Capability inventory: The skill can execute multiple shell commands via the
rwutility, which acts as a gateway to financial data tools. - Sanitization: The skill includes a 'Parameter Precheck' step that normalizes ticker symbols and validates enum values for specific fields.
- [EXTERNAL_DOWNLOADS]: The documentation provides instructions to configure the CLI using an external MCP gateway URL (
https://your-mcp-gateway.example.com).
Audit Metadata