LocalCode Middleware Server

A Python middleware server that sits between OpenCode and GLM 4.7, logging all requests and responses.

Features

• OpenAI-Compatible API: /v1/chat/completions endpoint
• Request Logging: Pretty prints all incoming requests
• Response Logging: Pretty prints all outgoing responses
• Streaming Support: Native Python async generators
• Tool Call Markers: Detects and marks tool requests in logs
• Functional Design: No OOP, pure functions

Installation

cd LocalCode
pip install -r requirements.txt

Running

python main.py

Server starts on http://0.0.0.0:4242

OpenCode Configuration

Add to OpenCode provider configuration:

{
  "localcode": {
    "npm": "@ai-sdk/openai-compatible",
    "api": "http://localhost:4242/v1",
    "name": "LocalCode",
    "models": {
      "glm-4.7": {
        "id": "glm-4.7",
        "name": "GLM-4.7 (via LocalCode)",
        "family": "glm-4.7",
        "reasoning": true,
        "tool_call": true,
        "temperature": true,
        "cost": {
          "input": 0,
          "output": 0
        },
        "limit": {
          "context": 204800,
          "output": 131072
        }
      }
    }
  }
}

Or configure via environment:

export OPENCODE_PROVIDER_URL="http://localhost:4242/v1"

API Endpoints

POST /v1/chat/completions

OpenAI-compatible chat completions endpoint.

Request Format:

{
  "model": "glm-4.7",
  "messages": [{ "role": "user", "content": "Hello" }],
  "stream": false,
  "temperature": 0.8,
  "max_tokens": 2000
}

Response Format:

{
  "id": "chatcmpl-xxx",
  "object": "chat.completion",
  "created": 1234567890,
  "model": "glm-4.7",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello!"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 10,
    "completion_tokens": 5,
    "total_tokens": 15
  }
}

Tool Call Example:

{
  "model": "glm-4.7",
  "messages": [{ "role": "user", "content": "Edit file" }],
  "tools": [
    {
      "type": "function",
      "function": {
        "name": "edit",
        "parameters": {
          "type": "object",
          "properties": {
            "filePath": { "type": "string" },
            "oldString": { "type": "string" },
            "newString": { "type": "string" }
          }
        }
      }
    }
  ]
}

GET /health

Health check endpoint.

Response:

{
  "status": "ok",
  "provider": "LocalCode",
  "model": "GLM-4.7"
}

Log Output

The middleware prints detailed logs to console:

Request Log

================================================================================
[REQUEST] 14:32:15
Model: glm-4.7
Stream: false
Messages count: 2
  [0] system: You are a coding agent...
  [1] user: Edit the function...
[Tool Definitions] 3 tools
  [0] edit: Edits a file...
================================================================================

Response Log

--------------------------------------------------------------------------------
[RESPONSE] 14:32:20
Content: Sure, I'll help you edit the function...
Finish reason: stop
Usage - prompt: 150, completion: 45, total: 195
--------------------------------------------------------------------------------

Tool Call Markers

[Tool Request] Detected tools in request
[Tool Call] edit
Args: {"filePath": "/src/app.ts", "oldString": "...", "newString": "..."}

Stream Chunks

[STREAM CHUNK] 14:32:18 Sure
[STREAM CHUNK] 14:32:18 , I'll
[STREAM CHUNK] 14:32:18  help

Configuration

Constants in main.py:

GLM_API_URL = "https://api.z.ai/api/coding/paas/v4"
PORT = 4242
API_KEY = "dummy"  # GLM 4.7 is free

Development

Running with custom port:

# Edit main.py
PORT = 8080

Verbose logging:

Logs are always printed to stdout. Redirect to file:

python main.py 2>&1 | tee middleware.log

Troubleshooting

Port already in use:

lsof -ti:4242 | xargs kill -9

Import errors:

pip install --upgrade pip
pip install -r requirements.txt

License

Same as parent OpenCode project.