gate-mcp-installer
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFECREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill clones the gate-skills library from the vendor's GitHub repository (https://github.com/gate/gate-skills.git) during the setup process.
- [COMMAND_EXECUTION]: The installer (install.sh) utilizes shell commands to detect local development environments, manage configuration paths, and handle repository cloning for Cursor, Claude, Codex, and OpenClaw.
- [REMOTE_CODE_EXECUTION]: Configures MCP server endpoints to execute vendor packages via npx (gate-mcp, gatepay-local-mcp), which downloads and runs code from the npm registry.
- [DYNAMIC_EXECUTION]: Employs a Node.js utility (merge-mcp-config.js) to dynamically update and merge MCP server definitions into the host environment's configuration files.
- [CREDENTIALS_UNSAFE]: The installation process prompts users for sensitive API keys (GATE_API_KEY, GATE_API_SECRET) and stores them in plain text within application configuration settings.
- Includes a static service API key (MCP_AK_8W2N7Q) in the DEX MCP fragments for server authentication.
- [INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection as it processes external repository data and user inputs during configuration.
- Ingestion points: Cloned repository content and user environment variables.
- Boundary markers: The agent is instructed to follow external runtime rules (gate-runtime-rules.md).
- Capability inventory: Includes file writes, shell execution, and network access via git/npx.
- Sanitization: Uses JSON parsing for configuration merging but does not perform independent validation of the external skill content.
Audit Metadata