mcp-management
Pass
Audited by Gen Agent Trust Hub on Oct 9, 2026
Risk Level: SAFECOMMAND_EXECUTIONREMOTE_CODE_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides a command-line interface (
scripts/cli.ts) and detailed instructions for executing shell commands to manage dependencies, list capabilities, and call MCP tools. - [REMOTE_CODE_EXECUTION]: The core functionality involves spawning subprocesses for MCP servers using the
@modelcontextprotocol/sdkandStdioClientTransport. These executions are driven by a user-controlled configuration file (.claude/.mcp.json) and typically target well-known servers from the official Model Context Protocol organization. - [EXTERNAL_DOWNLOADS]: The skill recommends the installation of the
gemini-cliand various MCP servers vianpmandnpx. These references target official package registries and well-known service providers. - [INDIRECT_PROMPT_INJECTION]: The skill establishes an attack surface for indirect prompt injection by ingesting data from external MCP tools (such as web search or browser automation) and utilizing a
GEMINI.mdfile to steer LLM behavior toward structured JSON outputs. This is a standard characteristic of tool-orchestration skills and is documented for performance and consistency.
Audit Metadata