A Python-based single-agent demo showing an execution loop, memory, context management, and tool integrations, with CLI setup and development workflow. It targets using MiniMax-M2.5 and supports Claude skills and MCP tools.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Mini Agent is a minimal yet professional demo project that showcases the best practices for building agents with the MiniMax M2.5 model. It leverages an Anthropic-compatible API and supports interleaved thinking for long, complex tasks.
How it works
Features include a Full Agent Execution Loop, Persistent Memory via a Session Note Tool, Intelligent Context Management with automatic conversation summarization, integration with Claude Skills (15 skills), MCP Tool integration for knowledge graph access and web search, and comprehensive logging. The project includes a CLI and a codebase intended to be easy to understand for building advanced agents.
Getting started
Quick Start steps include acquiring an API key for Global or China platforms, installing uv, and choosing a usage mode:
- Quick Start Mode installation commands:
# 1. Install directly from GitHub uv tool install git+https://github.com/MiniMax-AI/Mini-Agent.git # 2. Run setup script (creates config files) # macOS/Linux: curl -fsSL https://raw.githubusercontent.com/MiniMax-AI/Mini-Agent/main/scripts/setup-config.sh | bash # Windows (PowerShell): Invoke-WebRequest -Uri "https://raw.githubusercontent.com/MiniMax-AI/Mini-Agent/main/scripts/setup-config.ps1" -OutFile "$env:TEMP\setup-config.ps1" powershell -ExecutionPolicy Bypass -File "$env:TEMP\setup-config.ps1"
Configuration is stored in ~/.mini-agent/config/config.yaml and must include api_key, api_base, and model. Example:
api_key: "YOUR_API_KEY_HERE"
api_base: "https://api.minimax.io"
# api_base: "https://api.minimaxi.com" # China
model: "MiniMax-M2.5"
Run methods:
# Run as module for debugging
uv run python -m mini_agent.cli
# Install in editable mode for development
uv tool install -e .
# After installation, run from anywhere
mini-agent
Development guidance points to syncing dependencies with uv and optional config initialization steps, including cloning the repo, installing uv, syncing dependencies, and initializing Claude Skills. There are also references to Development Guide and Production Guide in the docs.
Recent releases
No releases listed in the latest release entry.
Traction
GitHub stars: 2934 Forks: 432 Open issues: 46
Behind the repo
Not provided in the README.
Caveats
License: MIT Created: 2025-10-31 Last push: 2026-02-14
License and basic project metadata are specified; troubleshooting includes SSL certificate error fixes and module not found guidance.