AgentAPI

AgentAPI is a lightweight framework for building agentic AI backends with a familiar Python/FastAPI style.

What You Get

• A stateful Agent with memory and tool calling
• Provider abstraction for OpenAI, Gemini, and OpenRouter
• Simple web API layer using AgentApp
• Built-in streaming support through Server-Sent Events (SSE)
• CLI commands to scaffold and run projects

Core Concepts

Agent

Agent handles prompt orchestration, model calls, tool execution, and conversation memory.

AgentApp

AgentApp extends FastAPI with chat-focused decorators:

• @app.chat for regular or streaming responses
• @app.stream as a streaming-only compatibility alias

@tool

Mark plain Python functions as LLM-callable tools. AgentAPI builds provider-compatible tool schemas automatically.

Quick Example

from agentapi import Agent, AgentApp

app = AgentApp()

agent = Agent(
    system_prompt="You are a helpful assistant",
    provider="openai",
)


@app.chat("/chat")
async def chat(message: str):
    return await agent.run(message)


@app.chat("/stream")
async def stream_chat(message: str):
    return agent.stream(message)

Next Steps

1. Start with Installation
2. Follow Getting Started
3. Configure Providers
4. Add Tool Calling