minimal-harness 0.4.4

An exploration of making an agent sdk as lean as possible while being effective.

minimal-harness

Documentation: /docs

A lightweight Python agent harness for building LLM-powered agents with tool-calling support.

Latest version: 0.4.4

What This Project Is For

Minimal-harness is a lean framework for building agents that can call tools. It provides:

• OpenAI-compatible API - Works with any OpenAI-compatible API provider
• Tool system - Create tools via decorators; includes built-in tools (bash, file ops)
• AsyncIterator events - Real-time async iteration for chunks, tool start/end, execution events
• Conversation memory - Tracks token usage across interactions
• ESC stop support - Gracefully stop LLM streaming and tool execution

Architecture

The framework uses an event-driven architecture with AsyncIterator-based event handling:

Agent (OpenAIAgent) → Internal Events → to_client_event() → Client-Facing Events

Event flow:

async for event in agent.run(user_input=[{"type": "text", "text": "..."}]):
    if isinstance(event, LLMChunk):
        # handle chunk
    elif isinstance(event, ToolEnd):
        # handle tool result

How to Build an App

Project Structure

A typical app looks like this:

my-app/
├── cli.py          # Entry point
└── tools.py        # Your custom tools

1. Create Your Entry Point

import argparse
import os
from openai import AsyncOpenAI
from minimal_harness.agent.openai import OpenAIAgent
from minimal_harness.client.events import (
    AgentStartEvent,
    AgentEndEvent,
    LLMChunkEvent,
    ToolStartEvent,
    ToolEndEvent,
)
from minimal_harness.llm.openai import OpenAILLMProvider
from minimal_harness.memory import ConversationMemory
from minimal_harness.tool.built_in.bash import get_tools as get_bash_tools

def main():
    parser = argparse.ArgumentParser(description="My AI agent")
    parser.add_argument("--base-url", required=True)
    parser.add_argument("--api-key", required=True)
    parser.add_argument("--model", default="qwen3.5-27b")
    args = parser.parse_args()

    client = AsyncOpenAI(base_url=args.base_url, api_key=args.api_key)
    llm_provider = OpenAILLMProvider(client=client, model=args.model)
    memory = ConversationMemory(system_prompt="You are a helpful assistant.")
    agent = OpenAIAgent(
        llm_provider=llm_provider,
        tools=list(get_bash_tools().values()),
        memory=memory,
    )

    async def run():
        stop_event = asyncio.Event()
        async for event in agent.run(
            user_input=[{"type": "text", "text": "What files are in the current directory?"}],
            stop_event=stop_event,
        ):
            client_event = event.to_client_event()
            if isinstance(client_event, AgentStartEvent):
                print(f"Agent starting...")
            elif isinstance(client_event, LLMChunkEvent):
                chunk = client_event.chunk
                if chunk and chunk.choices:
                    content = chunk.choices[0].delta.content or ""
                    print(content, end="", flush=True)
            elif isinstance(client_event, ToolStartEvent):
                print(f"\n[Calling tool: {client_event.tool_call['function']['name']}]")
            elif isinstance(client_event, ToolEndEvent):
                print(f"\n[Tool result: {client_event.result[:100]}...]")
            elif isinstance(client_event, AgentEndEvent):
                break

    import asyncio
    asyncio.run(run())

if __name__ == "__main__":
    main()

2. Add Custom Tools

Use the @register_tool decorator to add your own tools. You need a ToolRegistry instance:

from typing import AsyncIterator

from minimal_harness.tool.registration import register_tool
from minimal_harness.tool.registry import ToolRegistry

registry = ToolRegistry()

@register_tool(
    name="get_weather",
    description="Get weather for a location",
    parameters={
        "type": "object",
        "properties": {"location": {"type": "string"}},
        "required": ["location"],
    },
    registry=registry,
)
async def get_weather(location: str) -> AsyncIterator[dict]:
    yield {"success": True, "result": f"The weather in {location} is sunny."}

The decorator registers the tool with the provided registry. Pass the same registry to the harness when running.

3. Run

python cli.py --base-url https://api.openai.com/v1 --api-key sk-... --model gpt-4o

Or set environment variables:

export MH_BASE_URL=https://api.openai.com/v1
export MH_API_KEY=sk-...
export MH_MODEL=gpt-4o
python cli.py

Built-in Tools

Tool	Description
bash	Execute shell commands with timeout
local_file_operation	Read, write, patch, or delete files (4 universal modes)

Event Types

Event	Description
AgentStartEvent	Agent execution started
AgentEndEvent	Agent execution completed
LLMStartEvent	LLM generation started
LLMChunkEvent	LLM output chunk received
LLMEndEvent	LLM generation completed
ExecutionStartEvent	Tool execution started
ExecutionEndEvent	Tool execution completed
ToolStartEvent	Tool call started
ToolProgressEvent	Tool intermediate progress
ToolEndEvent	Tool call completed with result

Environment Variables

Variable	Description
MH_BASE_URL	API base URL
MH_API_KEY	API key
MH_MODEL	Model name (default: qwen3.5-27b)

Stop Mechanism

Press ESC during execution to gracefully stop LLM streaming and tool execution.