Installation & Setup

JetBrains Marketplace

Install directly from JetBrains Marketplace: AutoDev Sketch

Note: The plugin is named AutoDev Sketch in JetBrains Marketplace with ID com.unitmesh.autodev. This differs from custom repository and GitHub release versions.

This version works with IDE 2024.1 and newer (241+).

Custom Repository

1. Go to Settings → Plugins → Marketplace → Manage Plugin Repositories
2. Add the following URL:

https://plugin.unitmesh.cc/updatePlugins.xml

GitHub Releases

1. Download appropriate version from GitHub Releases
   • AutoDev-*-222.zip — For 2022.2 to 2023.2 versions
   • AutoDev-*-233.zip — For 2023.3 version
   • AutoDev-*-241.zip — For 2024.1 and newer versions
2. Install plugin from disk in JetBrains IDE

Configuration

After installation, configure the plugin in Settings → Tools → AutoDev

Default LLM

Supported providers: Deepseek (#96), Moonshot AI, ChatGLM(#90)

Basic Configuration

1. Open AutoDev configuration in Settings → Tools → AutoDev
2. Configure LLM Server URL, e.g.:
   • Deepseek: https://api.deepseek.com/chat/completions
   • OpenAI: https://api.openai.com/v1/chat/completions
3. Enter your LLM Key (API Key)
4. Set Custom Response Format using JsonPath, e.g.:
   • $.choices[0].delta.content
5. Configure Custom Request Format, e.g.:
   • { "customFields": {"model": "deepseek-chat", "stream": true }}

Advanced Configuration

For more detailed configuration options, see Custom LLM Server

Additional Models

Available model types:

• Plan: For reasoning and planning (Recommended: DeepSeek R1)
• Completion: For code completion
• FastApply: For fix patch generation (e.g., Kortix/FastApply-1.5B-v1.0)
• Others: General placeholder
• Act: (Not ready) For action execution (e.g., DeepSeek V3, Qwen 72B)
• Embedding: (Not ready) For embedding functions (e.g., sentence-transformers/all-MiniLM-L6-v2)

DeepSeek R1

{
  "name": "DeepSeek R1",
  "url": "https://api.deepseek.com/chat/completions",
  "auth": {
    "type": "Bearer",
    "token": "sk-ii"
  },
  "requestFormat": "{ \"customFields\": {\"model\": \"deepseek-reasoner\", \"stream\": true}}",
  "responseFormat": "$.choices[0].delta.content",
  "modelType": "Plan"
}

GLM4-Plus

{
  "name": "GLM4-Plus",
  "url": "https://open.bigmodel.cn/api/paas/v4/chat/completions",
  "auth": {
    "type": "Bearer",
    "token": "sk-ii"
  },
  "requestFormat": "{ \"customFields\": {\"model\": \"glm-4-plus\", \"stream\": true}}",
  "responseFormat": "$.choices[0].delta.content",
  "modelType": "FastApply"
}

DifyAI

{
   "name": "DifyAI",
   "description": "Dify Example",
   "url": "https://api.dify.ai/v1/completion-messages",
   "auth": {
      "type": "Bearer",
      "token": "app-abcd"
   },
   "requestFormat": "{\"fields\": {\"inputs\": {\"feature\": \"$content\"}, \"response_mode\": \"streaming\", \"user\": \"phodal\" }}",
   "responseFormat": "$.answer",
   "modelType": "Others"
}

Configuration fields:

• URL: LLM server address containing endpoint path
• Auth: Authentication information (currently only supports Bearer token)
• RequestFormat: JSON structure for API requests
• ResponseFormat: JsonPath for extracting content from response
• ModelType: Model type (see list above)