remote-mcp-server-cf
This project implements a Model Context Protocol (MCP) server for the COMPLiQ platform using Cloudflare Workers. It allows AI assistants to interact with COMPLiQ's API, logging various interactions such as prompts and files. Key features include prompt submission, file attachment, and result processing.
COMPLiQ MCP Server on Cloudflare
This project implements a Model Context Protocol (MCP) server for COMPLiQ on Cloudflare Workers. The server enables AI assistants to interact with the COMPLiQ platform through MCP tools.
What is COMPLiQ?
COMPLiQ is a platform that logs prompts, attached files, intermediate results, and responses from user interactions with AI models. It provides a standardized API for recording these interactions.
Features
The MCP server provides the following tools that correspond to COMPLiQ's API endpoints:
- inputPrompt - Submit a prompt/request to COMPLiQ (Mandatory)
- addFile - Attach a file to a request (Optional)
- intermediateResults - Send intermediate processing results (Optional)
- processingResult - Submit the final processing result/answer (Mandatory)
Setup
- Clone this repository
- Configure your API key in
wrangler.jsonc
or using environment secrets:npx wrangler secret put COMPLIQ_API_KEY
- Deploy the worker:
npm run deploy
Connecting to the MCP Server
Once deployed, your MCP server will be available at:
https://compliq-mcp-server.<your-account>.workers.dev/sse
You can connect to it using an MCP client like Claude Desktop by adding this configuration:
{
"mcpServers": {
"compliq": {
"command": "npx",
"args": [
"mcp-remote",
"https://compliq-mcp-server.<your-account>.workers.dev/sse"
]
}
}
}
API Endpoints
The COMPLiQ API endpoints used by this server are:
https://ai-stage-be.compliq.io/v1/actions/task-input
https://ai-stage-be.compliq.io/v1/actions/file-input
https://ai-stage-be.compliq.io/v1/actions/resources-used
https://ai-stage-be.compliq.io/v1/actions/output
Developer Documentation
For more information about the COMPLiQ API, refer to the API documentation.
For more details about the Model Context Protocol, visit modelcontextprotocol.io.