MCP SuperAssistant is a TypeScript Chrome extension that integrates MCP tools with multiple AI platforms, enabling tool detection, execution, and result insertion from a local proxy setup. It supports various platforms and provides setup for connecting to a local MCP proxy server.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
MCP SuperAssistant Chrome Extension integrates the Model Context Protocol (MCP) tools with multiple AI platforms such as ChatGPT, Perplexity, Grok, Gemini, Google AI Studio, OpenRouter, DeepSeek, T3 Chat, GitHub Copilot, Mistral AI, Kimi, Qwen Chat, Z Chat and more.
How it works
The extension detects MCP tool calls in AI responses, executes MCP tools with a single click, and inserts tool results back into the AI conversation. It can render function calls and results, supports Auto-Execute, Auto-Submit, and Push Content modes, and persists preferences like sidebar position and theme. A flow diagram is described: AI chat interface generates tool calls, extension detects calls, sends via SSE to a local proxy server, which forwards to the actual MCP server and returns results to insert back into chat.
A local proxy connection is established by running a MCP SuperAssistant Proxy via npx with a config.json that defines mcpServers. Example commands include starting the proxy with different output transport methods (sse, streamableHttp, ws) and a --help view for options. The steps include starting the proxy, opening the MCP sidebar, connecting to a local server URL (default http://localhost:3006/sse), and selecting the transport method.
Getting started
Key prerequisites include a local MCP server configuration (config.json) and running the proxy via one of:
npx -y @srbhptl39/mcp-superassistant-proxy@latest --config ./config.json --outputTransport sse
npx -y @srbhptl39/mcp-superassistant-proxy --config ./config.json --outputTransport streamableHttp
npx -y @srbhptl39/mcp-superassistant-proxy --config ./config.json --outputTransport ws
Usage involves:
- Opening the MCP SuperAssistant sidebar in a supported AI platform
- Configuring MCP Tools to enable/disable tools
- Attaching the MCP instructions prompt into the chat
- Executing tool calls manually or via Auto-Execute mode
- Using Auto-Submit and Push Content modes as needed
Recent releases
Latest release: v0.6.0 (2026-02-09) with fixes for Qwen MCP button visibility and Perplexity UI updates, and a downgrade of a model context SDK dependency to 1.20.2. Earlier notable updates include v0.5.9-alpha (2026-01-17) upgrading @modelcontextprotocol/sdk to 1.25.2 and zod to 4.3.5, and v0.5.8-alpha (2025-11-03) with a ChatGPT function name and call id fix. The changelog also mentions v0.5.7-alpha and v0.5.6-alpha releases with SDK upgrades and interaction improvements.
Traction
Stars: 2539; Forks: 343; Open issues: 97
License
MIT
Getting started (commands)
Install and start the development workflow commands are provided in the Development section:
# Install dependencies
pnpm install
# Start development server
pnpm dev
# Build for production
pnpm build
# Create zip package for distribution
pnpm zip