VM0Agent Skills
DeepSeek
AI language model API for chat and code generation
DeepSeek provides powerful AI language models for chat completion, reasoning, and code generation with OpenAI-compatible endpoints.
Required Environment
| Name | Type | Description |
|---|---|---|
DEEPSEEK_API_KEY | secret | API key from DeepSeek Platform |
Configuration
version: "1.0"
agents:
my-agent:
framework: claude-code
skills:
- https://github.com/vm0-ai/vm0-skills/tree/main/deepseekRun
Store your secret on the platform (recommended, one-time setup):
vm0 secret set DEEPSEEK_API_KEY your-deepseek-api-keyThen run your agent - secret is automatically loaded:
vm0 run my-agent "generate code using DeepSeek"For CI/CD or temporary overrides, pass secrets at runtime: --secrets DEEPSEEK_API_KEY=value. See Environment Variables for details.
Example Instructions
# Code Assistant Agent
You use DeepSeek API for code generation tasks.
## Workflow
1. Receive coding request from user
2. Call DeepSeek chat completion API
3. Process and format the response
4. Return generated code
## Models
- deepseek-chat: General purpose
- deepseek-reasoner: Complex reasoning# Reasoning Agent
You use DeepSeek reasoner for complex problem solving.
## Workflow
1. Receive complex problem
2. Use deepseek-reasoner model
3. Analyze step-by-step reasoning
4. Present solution with explanation
## Guidelines
- Use reasoner for math and logic
- Include temperature control for creativity