Skip to main content
Superserve works with any agent framework. Whether you’re using Claude Agent SDK, OpenAI Agents SDK, LangChain, Mastra, Pydantic AI, or a custom implementation, you can deploy to Superserve with one command.

​
Supported Frameworks

Claude Agent SDK

Anthropic’s official SDK for building agents with Claude

OpenAI Agents SDK

OpenAI’s Python SDK for building agentic applications

LangChain

The popular framework for building LLM applications

Mastra

TypeScript framework for building AI agents

Pydantic AI

Type-safe agent framework built on Pydantic

Custom Framework

Use stdin/stdout for any custom implementation

​
How It Works

Superserve detects your framework automatically by analyzing your code and dependencies. The deployment process is the same regardless of framework:
1

Write your agent

Build your agent using your preferred framework. Follow the framework’s own conventions and best practices.
2

Deploy to Superserve

Run superserve deploy agent.py (or agent.ts for TypeScript frameworks). Superserve analyzes dependencies and builds a container image.
3

Set secrets

Configure API keys with superserve secrets set my-agent API_KEY=.... Secrets are encrypted and injected securely at runtime.
4

Run your agent

Start an interactive session with superserve run my-agent. Your agent runs in an isolated Firecracker microVM with persistent storage.

​
Framework Requirements

All frameworks work with Superserve as long as they:
  • Accept input from stdin or expose a simple run loop
  • Output responses to stdout
  • Can be packaged as a Python or Node.js application
Superserve automatically handles:
  • Dependency installation (pip, npm, bun, etc.)
  • Environment variable injection
  • Session persistence
  • Isolation and security

​
Custom Frameworks

If you’re using a custom framework or building your own agent from scratch, you can still deploy to Superserve. The only requirement is a simple stdin/stdout interface:
agent.py
import sys

while True:
    try:
        user_input = input()
    except EOFError:
        break
    
    # Process input with your custom logic
    response = your_agent_function(user_input)
    print(response)
Deploy it like any other agent: 
superserve deploy agent.py --name my-custom-agent
superserve run my-custom-agent

​
Language Support

Superserve supports agents written in:
  • Python - Detected via .py extension or pyproject.toml/requirements.txt
  • TypeScript/JavaScript - Detected via .ts/.js extension or package.json
Dependencies are installed automatically:
  • Python: pip install from requirements.txt or pyproject.toml
  • Node.js: npm install or bun install from package.json

​
Next Steps

Explore the framework-specific guides to see complete working examples:

Claude Agent SDK

OpenAI Agents SDK

LangChain

Mastra

Pydantic AI

Build docs developers (and LLMs) love

Get started for freeTalk to us