mcp-client
by otoTree
The `mcp-client` project exposes a local mcp-server through Flask, allowing users to create their own intranet penetration for interaction between platforms like Coze/Dify/FastGPT and local resources.
Last updated: N/A
What is mcp-client?
mcp-client is a project that exposes a local mcp-server through a Flask application, enabling users to create their own intranet penetration. This allows platforms like Coze, Dify, and FastGPT to interact with resources on the user's local network.
How to use mcp-client?
To use mcp-client, you need to set up the environment by installing uv, creating a virtual environment, and installing dependencies from the pyproject.toml file. You then configure the server information in mcp_server.json and environment variables in .env. Finally, run the server.py file to start the server and use the provided API endpoints to interact with the exposed services.
Key features of mcp-client
Exposes local services through a Flask API
Supports custom server configurations via
mcp_server.jsonUses environment variables for sensitive information
Provides API endpoints for listing and calling tools on the server
Use cases of mcp-client
Integrating local tools with AI platforms like Coze, Dify, and FastGPT
Creating a secure tunnel to access local resources from remote locations
Developing and testing AI agents that require access to local data
Building custom integrations between local systems and cloud services
FAQ from mcp-client
How do I configure the services that are exposed?
How do I configure the services that are exposed?
You configure the services in the mcp_server.json file. Each service definition includes a description, command, and arguments.
How do I protect sensitive information?
How do I protect sensitive information?
Sensitive information such as API keys should be stored in the .env file and added to .gitignore to prevent it from being committed to version control.
What is the purpose of the virtual environment?
What is the purpose of the virtual environment?
The virtual environment isolates the project's dependencies, preventing conflicts with other projects.
How do I call a specific tool on a server?
How do I call a specific tool on a server?
You send a POST request to the /call_tool endpoint with the server name, tool name, arguments, and secret key in the request body.
Why am I getting certificate errors on cloud platforms?
Why am I getting certificate errors on cloud platforms?
The project uses a self-signed certificate, which may not be trusted by cloud platforms. Consider using a certificate from a trusted CA for production deployments.