summon-app
Postman for MCP servers
claude mcp add --transport stdio twillai-summon-app node apps/electron/main.js
How to use
Summon is a desktop client designed to help teams build, test, and manage MCP servers and AI agents. It provides a visual workflow for importing API definitions, configuring authentication (including mock options), and discovering tools automatically from MCP servers. The app also includes an Interactive AI Playground to experiment with tool calls across multiple AI models, live tool editing to adjust tool names, descriptions, and input schemas on the fly, and support for creating datasets and evaluations to gauge tool calling accuracy. In practice, you would use Summon to connect to existing MCP servers or to generate new MCP implementations from OpenAPI specs, then test and iterate on tool usage within the integrated environment. The ecosystem supports external MCPs via mcp.json configuration and aims to streamline tool discovery and evaluation without leaving the desktop experience.
How to install
Prerequisites:
- A modern operating system (macOS, Windows, or Linux)
- Node.js and npm installed on your system (for building from source or running development scripts)
- Git installed for cloning the repository
From Source:
-
Clone the repository git clone https://github.com/TrySummon/summon-app.git cd summon-app
-
Install dependencies npm install
-
Run the development server (Summon desktop client) npm start
-
Build for your platform (optional) npm run make
See package.json for other build scripts
Note: As a desktop Electron application, Summon bundles MCP-related tooling and a local runtime to interact with MCP servers. If you simply want to use the prebuilt application, download the latest release for your OS from the provided link in the README and install it per your OS conventions.
Additional notes
Tips and considerations:
- The Summon app uses the Model Context Protocol SDK to interact with MCP servers and AI agents; ensure your network policies allow local/remote MCP interactions as needed.
- When configuring MCP connections, you can leverage OpenAPI imports, API authentication options, and mock servers to test tool calls without relying on external services.
- The Live Tool Editing feature enables you to modify tool metadata and input schemas without restarting servers, with real-time diff views to track changes.
- If you encounter issues with local MCP discovery, verify that the target MCP server exposes the expected endpoints and that the Summon app has the necessary permissions to reach them (local firewall rules, etc.).
- Check the roadmap and release notes for updates to MCP SDKs, authentication methods, and new tooling options (e.g., new AI models or evaluation workflows).
Related MCP Servers
cipher
Byterover Cipher is an opensource memory layer specifically designed for coding agents. Compatible with Cursor, Codex, Claude Code, Windsurf, Cline, Claude Desktop, Gemini CLI, AWS's Kiro, VS Code, Roo Code, Trae, Amp Code and Warp through MCP. Built by https://byterover.dev/
HyperChat
HyperChat is a Chat client that strives for openness, utilizing APIs from various LLMs to achieve the best Chat experience, as well as implementing productivity tools through the MCP protocol.
aser
Aser is a lightweight, self-assembling AI Agent frame.
mindbridge
MindBridge is an AI orchestration MCP server that lets any app talk to any LLM — OpenAI, Anthropic, DeepSeek, Ollama, and more — through a single unified API. Route queries, compare models, get second opinions, and build smarter multi-LLM workflows.
codemesh
The Self-Improving MCP Server - Agents write code to orchestrate multiple MCP servers with intelligent TypeScript execution and auto-augmentation
cco
Real-time audit and approval system for Claude Code tool calls.