mcp-cli
Pass
Audited by Gen Agent Trust Hub on Sep 20, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill enables an agent to interact with external tools and data sources via MCP servers, facilitating the ingestion of untrusted content into the agent's execution context. This creates a surface for potential injection attacks where retrieved data could influence subsequent agent behavior. * Ingestion points: Data and tool outputs retrieved from MCP servers (SKILL.md). * Boundary markers: No explicit markers or warnings are provided to handle untrusted tool output. * Capability inventory: The skill enables filesystem search/read and arbitrary API interactions. * Sanitization: No instructions for validation or sanitization of tool output are included.
- [COMMAND_EXECUTION]: The documentation includes examples of building shell command chains using pipes, xargs, and sh to process results from mcp-cli. This demonstrates a pattern for executing system commands based on data retrieved from external, potentially untrusted sources.
- [NO_CODE]: This skill is entirely comprised of markdown documentation and contains no scripts or executable code files.
Audit Metadata