mcp-tooling-fit
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze repository facts—including manifests, frameworks, database configurations, and CI/deployment files—to determine which tools to enable. This introduces a surface where malicious content in these untrusted files could potentially influence the agent's tool selection or behavior.
- Ingestion points: Repository manifests, framework files, DB/config files, and CI/deployment scripts as specified in Step 2 of SKILL.md.
- Boundary markers: The instructions do not define clear boundaries or delimiters to separate analyzed data from agent instructions.
- Capability inventory: The agent can activate or deactivate tools using the
agentitCLI and persist state to the local.agentit/STATE.mdfile. - Sanitization: There are no instructions for sanitizing or validating the content read from the repository files before it is processed.
- [COMMAND_EXECUTION]: The skill utilizes the
agentitCLI tool for inventory management and state inspection (e.g.,agentit mcp status,agentit mcp active).
Audit Metadata