MiroFlow is an open-source Python-based agent framework focusing on multi-step internet research. It includes MiroFlow, MiroThinker, and MiroVerse data, with benchmarks and demo resources available.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
This repository is the official implementation of the MiroMind Research Agent Project. It includes the MiroFlow framework, MiroThinker, and MiroVerse data. The project targets multi-step internet research tasks such as future event prediction and provides demo and benchmark materials.
How it works
MiroFlow is a modular framework for building AI agents that perform research tasks. It supports multiple models and tools, including components like Audio Transcription, Python tools, File Reading, Reasoning, and Google Search, among others. Documentation references indicate evaluation across multiple benchmarks (FutureX, GAIA, HLE, xBench-DeepSearch, BrowserComp).
Getting started
Prerequisites
- Python 3.12 or higher
- Package Manager uv
- Linux, macOS
Quick Setup
# 1. Clone and setup
git clone https://github.com/MiroMindAI/MiroFlow && cd MiroFlow
uv sync
# 2. Configure API key
cp .env.template .env
# Edit .env and add your OPENROUTER_API_KEY
# 3. Run your first agent
uv run main.py trace --config_file_name=agent_quickstart_reading --task="What is the first country listed in the XLSX file that have names starting with Co?" --task_file_name="data/FSI-2023-DOWNLOAD.xlsx"
Expected Output: Your agent should return Congo Democratic Republic
Tip: Ensure the API key is set in the .env file and dependencies are installed.
Recent releases
- None
Traction
- Stars: 3086
Behind the repo
- The README mentions partnerships with MiroThinker and MiroVerse data; no external startup/company links are provided in the excerpts.
Caveats
- License: Apache-2.0
- Created: 2025-08-05
- Last push: 2026-07-06
- Open issues: 2