DeepSeek V4 for Openclaw

A high-performance CLI tool to interact with DeepSeek V4 Flash and Pro models using an OpenAI-compatible API.

jiajiaoy
v1.0.1
Apr 28, 2026
0
802
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install deepseek-v4

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install deepseek-v4 using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).

Prefer to download?

Get the raw skill files in a ZIP archive.

What is DeepSeek V4?

This skill enables developers to leverage the powerful DeepSeek V4 models directly from their terminal. By providing a bridge to both the cost-efficient Flash model and the high-reasoning Pro model, it allows for seamless AI integration into local workflows. It is a standout addition to the Openclaw Skills library for those needing fast, reliable, and affordable AI assistance.

DeepSeek V4 offers a massive 1M token context and specialized thinking modes for complex reasoning. This implementation focuses on speed and simplicity, requiring no heavy dependencies other than a lightweight Python environment, making it perfect for developers who live in the terminal.

DeepSeek V4 Use Cases

  • Running one-shot Q&A for quick code explanations or text summaries.
  • Engaging in multi-turn chats for complex project brainstorming and debugging.
  • Using thinking mode to visualize the model's internal reasoning trace for math or logic problems.
  • Integrating fast, low-cost AI capabilities into existing shell scripts and automated pipelines.

How DeepSeek V4 Works

  1. The user authenticates by providing a DeepSeek API key through a standard environment variable.
  2. Commands are executed through specialized Python scripts managed by the uv package manager for high performance.
  3. The skill communicates with the DeepSeek API using a standard OpenAI-compatible protocol.
  4. Depending on the flags used (such as --think), the skill handles the streaming output of both reasoning tokens and the final response content.

DeepSeek V4 Setup

To get started with this skill from the Openclaw Skills collection, follow these steps:

  1. Obtain your API key from the DeepSeek Platform.
  2. Set the environment variable in your terminal:
export DEEPSEEK_API_KEY=your_key_here
  1. Ensure you have uv installed to run the scripts efficiently:
brew install uv
  1. Test the connection with a simple prompt:
uv run scripts/ask.py "What is the capital of France?"

DeepSeek V4 Data Schema & Taxonomy

Component Description
scripts/ask.py Handles single-turn queries and logic-heavy reasoning tasks.
scripts/chat.py Manages interactive sessions for persistent multi-turn conversations.
scripts/models.py Utility script to retrieve current model lists and pricing data.
API Schema Uses standard OpenAI-compatible JSON payloads for requests.

DeepSeek V4 Advanced Features

  • Thinking mode support for Pro models to expose the chain-of-thought reasoning process.
  • Customizable system prompts via the --system flag to enforce specific tone or constraints.
  • Toggleable streaming support with the --no-stream flag for cleaner integration into parent scripts.
  • Native support for the latest V4 architectures including Flash and Pro optimizations.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*