Awesome DeepSeek Agent

repository·main·Indexed 26 days ago

https://github.com/deepseek-ai/awesome-deepseek-agent

A curated collection of integration guides for using DeepSeek models, such as DeepSeek-V4-Pro and DeepSeek-V4-Flash, within AI agent frameworks, terminal-based coding assistants, and IDE extensions. Includes setup instructions for tools like AstrBot, Cherry Studio, Claude Code, Cline, and various CLI agents.

Tokens: 23.4K
Snippets: 66
Records: 145
Agent score: 88%

What's inside awesome-deepseek-agent

Overview of Awesome DeepSeek Agent
main
Awesome DeepSeek Agent is a curated collection of guides designed to help developers integrate DeepSeek models (such as DeepSeek-V4-Pro or DeepSeek-V4-Flash) into various AI agent frameworks and coding assistants. Each guide provides specific instructions for installation, configuration, and initial execution within the respective tool.
https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/README.md
Run Pi and select a DeepSeek model
main
To start using Pi with DeepSeek:
1. Navigate to your project directory: cd /path/to/my-project
2. Launch Pi by running the pi command.
3. Inside the Pi terminal, type /model to open the model switcher.
4. Select the deepseek provider and choose either DeepSeek-V4-Pro or DeepSeek-V4-Flash.
```
pi
```
https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/pi_mono.md
Run Codex with Moon Bridge
main
Once Moon Bridge is running and Codex is configured, you can use Codex in your project directory. Codex will route requests through Moon Bridge to DeepSeek.
Manual Start
1. Start Moon Bridge: go run ./cmd/moonbridge --config config.yml
2. In a new terminal, navigate to your project: cd /path/to/my-project
3. Run Codex: codex
One-Click Script
Use the provided scripts to automate the entire process (building/starting Moon Bridge, generating config, and starting Codex).
macOS / Linux:
```
./scripts/start_codex_with_moonbridge.sh --project-directory /path/to/my-project
```
Windows PowerShell:
```
.\scripts\start_codex_with_moonbridge.ps1 -ProjectDirectory C:\path\to\my-project
```
https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/codex.zh-CN.md

Configure local models in WorkBuddy/CodeBuddy

main

Create or edit a models.json file to define DeepSeek models.

File Locations:

User-level (Global): C:\Users\<your-username>\.codebuddy\models.json
Project-level: <your-project>\.codebuddy\models.json

Important Requirements:

Save the file as UTF-8 without BOM. Desktop versions may fail to read files with a UTF-8 BOM header.
Set your API key as an environment variable DEEPSEEK_API_KEY before configuring, or paste the key directly into the JSON.

Configuration Schema: Use the following structure to add deepseek-v4-pro and deepseek-v4-flash to your availableModels list.

{
  "models": [
    {
      "id": "deepseek-v4-pro",
      "name": "DeepSeek V4 Pro",
      "vendor": "DeepSeek",
      "url": "https://api.deepseek.com/v1/chat/completions",
      "apiKey": "${DEEPSEEK_API_KEY}",
      "maxInputTokens": 128000,
      "maxOutputTokens": 8192,
      "supportsToolCall": true,
      "supportsImages": false,
      "relatedModels": {
        "lite": "deepseek-v4-flash",
        "reasoning": "deepseek-v4-pro"
      }
    },
    {
      "id": "deepseek-v4-flash",
      "name": "DeepSeek V4 Flash",
      "vendor": "DeepSeek",
      "url": "https://api.deepseek.com/v1/chat/completions",
      "apiKey": "${DEEPSEEK_API_KEY}",
      "maxInputTokens": 128000,
      "maxOutputTokens": 8192,
      "supportsToolCall": true,
      "supportsImages": false
    }
  ],
  "availableModels": [
    "deepseek-v4-pro",
    "deepseek-v4-flash"
  ]
}

Configure GitHub Copilot CLI to use DeepSeek V4 models

main

To use DeepSeek V4 models with GitHub Copilot CLI via BYOK (Bring Your Own Key), you must use the anthropic provider type. Using the openai provider type will result in a 400 error because DeepSeek requires reasoning_content to be echoed back, which the OpenAI integration does not support. The Anthropic Messages API endpoint is used to avoid this.

Environment Variables (Linux / Mac)

export COPILOT_PROVIDER_TYPE=anthropic
export COPILOT_PROVIDER_BASE_URL=https://api.deepseek.com/anthropic
export COPILOT_PROVIDER_API_KEY=sk-your-deepseek-api-key
export COPILOT_MODEL=deepseek-v4-pro

Environment Variables (Windows PowerShell)

$env:COPILOT_PROVIDER_TYPE="anthropic"
$env:COPILOT_PROVIDER_BASE_URL="https://api.deepseek.com/anthropic"
$env:COPILOT_PROVIDER_API_KEY="sk-your-deepseek-api-key"
$env:COPILOT_MODEL="deepseek-v4-pro"

Available Models

deepseek-v4-pro
deepseek-v4-flash

# Linux / Mac example
export COPILOT_PROVIDER_TYPE=anthropic
export COPILOT_PROVIDER_BASE_URL=https://api.deepseek.com/anthropic
export COPILOT_PROVIDER_API_KEY=sk-your-deepseek-api-key
export COPILOT_MODEL=deepseek-v4-pro

Install Hermes Agent
main
You can install Hermes Agent on Linux, macOS, or WSL2 using a one-line installer. The only prerequisite is having git installed. The installer handles all other dependencies automatically.
```
curl -fsSL https://raw.githubusercontent.com/NousResearch/hermes-agent/main/scripts/install.sh | bash
```
https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/hermes.md
Configure Cline using OpenAI Compatible mode for DeepSeek
main
If you prefer using the OpenAI compatible interface to access DeepSeek, configure Cline with the following settings:
1. Select Bring my own API Key.
2. Set API Provider to OpenAI Compatible.
3. Set Base URL to https://api.deepseek.com.
4. Enter your DeepSeek API Key.
5. Enter the Model ID (e.g., deepseek-v4-pro).
6. (Optional) Use Model Configuration to adjust window size, temperature, pricing, and limits.
https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/cline.md

Configure Moon Bridge for DeepSeek

main

Moon Bridge acts as a forwarding layer between Codex and the DeepSeek API. After cloning the Moon Bridge repository, create a config.yml file. This configuration enables DeepSeek V4 Pro/Flash models and sets up the routing to use Moon Bridge as the default model for Codex.

mode: "Transform"

server:
  addr: "127.0.0.1:38440"

models:
  deepseek-v4-pro:
    context_window: 1000000
    max_output_tokens: 384000
    default_reasoning_level: "high"
    supported_reasoning_levels:
      - effort: "high"
        description: "High reasoning effort"
      - effort: "xhigh"
        description: "Extra high reasoning effort"
    supports_reasoning_summaries: true
    default_reasoning_summary: "auto"
    extensions:
      deepseek_v4:
        enabled: true
  deepseek-v4-flash:
    context_window: 1000000
    max_output_tokens: 384000
    default_reasoning_level: "high"
    supported_reasoning_levels:
      - effort: "high"
        description: "High reasoning effort"
      - effort: "xhigh"
        description: "Extra high reasoning effort"
    supports_reasoning_summaries: true
    default_reasoning_summary: "auto"
    extensions:
      deepseek_v4:
        enabled: true

providers:
  deepseek:
    base_url: "https://api.deepseek.com/anthropic"
    api_key: "sk-your-deepseek-api-key"
    offers:
      - model: deepseek-v4-pro
      - model: deepseek-v4-flash

routes:
  moonbridge:
    model: deepseek-v4-pro
    provider: deepseek

defaults:
  model: moonbridge
  max_tokens: 65536

Install Cherry Studio
main
Cherry Studio is an open-source desktop AI client for Windows, macOS, and Linux that allows unified access to various LLM providers.
Download the appropriate installer from the Cherry Studio Releases or the official website:
- Windows: .exe
- macOS: .dmg (supports Intel and Apple Silicon)
- Linux: .AppImage, .deb, or .rpm
https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/cherry_studio.zh-CN.md
Run Crush and switch to DeepSeek models
main
To start using Crush in your project:
1. Navigate to your project directory:
  cd /path/to/my-project
2. Launch Crush:
  crush
3. Switch Models:
  Press Ctrl+L or type /model to open the model switcher.
  Select the DeepSeek provider.
  Choose either DeepSeek-V4-Pro or DeepSeek-V4-Flash to begin coding.
```
crush
```
https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/crush.zh-CN.md
Use DeepSeek V4 for Advanced Features in Cherry Studio
main
Once configured, DeepSeek V4 can be utilized across several Cherry Studio features:
- Knowledge Bases: Use the Knowledge Base tab to upload PDFs, Markdown, or Office files for RAG (Retrieval-Augmented Generation) using DeepSeek V4.
- AI Translation: Select a DeepSeek V4 model as the engine on the Translate page.
- MCP Servers: Add servers via Settings → MCP to allow DeepSeek-driven conversations to call external tools.
- OpenClaw: Use the OpenClaw page in the sidebar to install the personal AI assistant and point it at your DeepSeek V4 model to power IM bots (e.g., for Feishu or WeChat).
https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/cherry_studio.md
Configure the DeepSeek Provider in LobeHub
main
To use DeepSeek models in LobeHub (Web, Desktop, or Self-hosted), you must configure the provider settings.
1. Navigate to Settings → Service Model or go directly to the DeepSeek provider page at https://app.lobehub.com/settings/provider/deepseek.
2. Select DeepSeek from the provider list and ensure the provider switch is enabled.
3. Enter your DeepSeek API Key (obtained from the DeepSeek Platform) into the API Key field.
4. Leave API Proxy URL empty unless using a custom proxy (default is https://api.deepseek.com).
5. Use Connectivity Check (defaults to deepseek-v4-flash) to verify the connection.
6. In the Model List, ensure DeepSeek V4 Pro and DeepSeek V4 Flash are enabled. Use Fetch models if they are missing.
https://github.com/deepseek-ai/awesome-deepseek-agent/blob/main/docs/lobehub.md

Awesome DeepSeek Agent

What's inside awesome-deepseek-agent

Manual Start

One-Click Script

Environment Variables (Linux / Mac)

Environment Variables (Windows PowerShell)

Available Models