Hello. Using Docker desktop on Mac with Cursor IDE. I have Archon started and the MCP started through streamlit UI. When trying to run archon, Cursor recognizes the MCP, but cannot connect. Here’s a screenshot for reference. Any help would be appreciated.
Waiting on this because I didn’t found a solution yet. It’s been 3 days of constant delete-reinstall. Asked AI for every possible solution but still no answer.
I think it’s a bug. Yesterday it was working but today, after uninstalling and reinstalling for five times it still doesn’t work. I even tried not using mpc via cursor and chatting via the streamlit ui but it always gives me an error.
I wish I could upload the images from my latest try but I can’t.
This is all that’s in the log of the MCP docker container:
2025-03-28 15:15:06 {“jsonrpc”:“2.0”,“id”:0,“result”:{“protocolVersion”:“2024-11-05”,“capabilities”:{“experimental”:{},“prompts”:{“listChanged”:false},“resources”:{“subscribe”:false,“listChanged”:false},“tools”:{“listChanged”:false}},“serverInfo”:{“name”:“archon”,“version”:“1.2.1”}}}
2025-03-28 15:15:07 {“jsonrpc”:“2.0”,“id”:1,“result”:{“tools”:[{“name”:“create_thread”,“description”:"Create a new conversation thread for Archon.\n Always call this tool before invoking Archon for the first time in a conversation.\n (if you don’t already have a thread ID)\n \n Returns:\n str: A unique thread ID for the conversation\n ",“inputSchema”:{“properties”:{},“title”:“create_threadArguments”,“type”:“object”}},{“name”:“run_agent”,“description”:"Run the Archon agent with user input.\n Only use this tool after you have called create_thread in this conversation to get a unique thread ID.\n If you already created a thread ID in this conversation, do not create another one. Reuse the same ID.\n After you receive the code from Archon, always implement it into the codebase unless asked not to.\n\n After using this tool and implementing the code it gave back, ask the user if they want you to refine the agent\n autonomously (they can just say ‘refine’) or they can just give feedback and you’ll improve the agent that way.\n\n If they want to refine the agent, just give ‘refine’ for user_input.\n \n Args:\n thread_id: The conversation thread ID\n user_input: The user’s message to process\n \n Returns:\n str: The agent’s response which generally includes the code for the agent\n ",“inputSchema”:{“properties”:{“thread_id”:{“title”:“Thread Id”,“type”:“string”},“user_input”:{“title”:“User Input”,“type”:“string”}},“required”:[“thread_id”,“user_input”],“title”:“run_agentArguments”,“type”:“object”}}]}}
