Gangtise MCP

简体中文 | English

Remote service: https://openapi.gangtise.com/application/mcp/ (served by this repository).

Get credentials · HTTP / SSE · CLI · Docker

Connect

1. Create an account on the open platform and copy Access Key / Secret Key from account settings.

2. Pick a connection mode. For remote use, point the client at the HTTP URL below and send AK/SK headers.

Examples below use GitHub: https://github.com/gangtise-ai/gangtise-data-channels. Chinese docs use Gitee — see README.cn.md.

Two auth methods can be used together. Server order: Authorization → AK/SK (accessKey / secretKey, or X-GTS-Credentials).

accessKey: <ak>
secretKey: <sk>

Authorization: Bearer <token>

Related MCP server: market-data-mcp

Install by platform

On clients with an agent, paste the JSON below and ask it to install. You can also write the same JSON into the client config file.

Remote HTTP (recommended). Write ~/.cursor/mcp.json or project .cursor/mcp.json:

{
  "mcpServers": {
    "gangtise": {
      "type": "streamableHttp",
      "url": "https://openapi.gangtise.com/application/mcp/",
      "headers": {
        "accessKey": "YOUR_ACCESS_KEY",
        "secretKey": "YOUR_SECRET_KEY"
      }
    }
  }
}

Open Cursor → Settings → Tools & MCP and enable gangtise. type must be streamableHttp. Do not point that type at /sse.

Local stdio (requires uv):

{
  "mcpServers": {
    "gangtise": {
      "command": "uvx",
      "args": [
        "--from",
        "git+https://github.com/gangtise-ai/gangtise-data-channels#subdirectory=mcp",
        "gangtise-mcp"
      ],
      "env": {
        "GTS_ACCESS_KEY": "YOUR_ACCESS_KEY",
        "GTS_SECRET_KEY": "YOUR_SECRET_KEY"
      }
    }
  }
}

Guide: WorkBuddy MCP. Use the same remote HTTP endpoint as Cursor: https://openapi.gangtise.com/application/mcp/, streamableHttp, AK/SK headers. After install, trust and enable it under Connectors → My MCP.

Edit claude_desktop_config.json with the same stdio JSON as Cursor (mcpServers.gangtise). Requires uv.

claude mcp add gangtise -- uvx \
  --from "git+https://github.com/gangtise-ai/gangtise-data-channels#subdirectory=mcp" \
  gangtise-mcp

Set GTS_ACCESS_KEY / GTS_SECRET_KEY on that MCP entry. Remote HTTP: docs/http-sse.md.

Workspace .vscode/mcp.json:

{
  "servers": {
    "gangtise": {
      "type": "http",
      "url": "https://openapi.gangtise.com/application/mcp/",
      "headers": {
        "accessKey": "YOUR_ACCESS_KEY",
        "secretKey": "YOUR_SECRET_KEY"
      }
    }
  }
}

HTTP  https://openapi.gangtise.com/application/mcp/
SSE   https://openapi.gangtise.com/application/mcp/sse

SSE type, X-GTS-Credentials, and header details: docs/http-sse.md. Self-host: docs/docker-deploy.md.

# stdio
cd mcp && uv sync
uv run gangtise-mcp

# HTTP, default 0.0.0.0:8000
uv run gangtise-mcp-api

# CLI
cd ../cli && uv sync
uv run gangtise list

After editing core/ or skills/openapi/, run python3 scripts/generate.py at the repo root before the commands above.

Chinese: README.cn.md.

License

MIT © 2026 Gangtise