mcp-gateway-configuration
Pass
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions for utilizing official packages from the @modelcontextprotocol organization and recommends pinning versions to prevent supply chain risks.
- [COMMAND_EXECUTION]: Documentation includes diagnostic shell commands intended for troubleshooting, such as verifying server availability with
which, running help commands withnpx, and checking environment variable configuration usingenv. - [INDIRECT_PROMPT_INJECTION]: The skill defines a configuration architecture where the agent interacts with external data from GitHub and the local filesystem, which constitutes a potential attack surface.
- Ingestion points: Data enters the agent context through the GitHub MCP server (code search, issues, PRs) and the Filesystem MCP server (local file reads) as described in SKILL.md.
- Boundary markers: The configuration examples do not define specific delimiters or instructions to ignore embedded prompts within ingested data.
- Capability inventory: The gateway configuration enables capabilities including file read/write operations and GitHub repository modifications (PR creation, code search).
- Sanitization: No specific data validation or sanitization requirements are specified for the gateway configuration level.
Audit Metadata