XiaomiMiMo/MiMo-Code

MiMo Code: Where Models and Agents Co-Evolve

What it solves

MiMoCode is a terminal-native AI coding assistant designed to handle complex, long-horizon development tasks. It addresses the problem of AI assistants losing context over long sessions, the high cost of long-context windows, and the the lack of structured, autonomous workflows for software engineering.

How it works

MiMoCode operates as a CLI tool that can read and write code, execute commands, and manage Git. It uses a persistent memory system (via SQLite FTS5) to maintain project knowledge and session state across restarts. To manage costs and performance, it employs intelligent context management, including automatic checkpoints and budgeted injection of memory and notes.

It utilizes a multi-agent system with specialized roles (Build, Plan, Compose) and can spawn subagents for parallel execution. It also supports "Workflows"—deterministic JavaScript scripts that orchestrate agents through fixed phases (e.g., Brainstorm → Design → Implement) without requiring user interaction.

Who it’s for

Developers who prefer a terminal-based workflow and want an AI agent that can autonomously handle feature delivery, deep research, and complex codebase exploration across multiple sessions.

Highlights

  • Persistent Memory: Uses MEMORY.md, checkpoint.md, and task logs to ensure the agent doesn't have to relearn the project context.
  • Multi-Agent Orchestration: Specialized agents for building, planning, and composing, with the ability to spawn parallel subagents.
  • Deterministic Workflows: Pre-built JS-based pipelines for full development cycles, deep research, and adversarial fact-checking.
  • Extensible Skill System: A library of reusable instruction sets for specific tasks like arXiv search, PDF/PPTX manipulation, and data analytics.
  • Goal-Based Stopping: An independent judge model evaluates if a set goal is truly satisfied before the agent stops working.
  • Voice Input: Integrated real-time streaming voice transcription for hands-free interaction.

Related

  • Project
    openJiuwen-ai/iCode

    iCode is a terminal-based AI agent orchestrator that provides a transparent, privacy-focused interface for managing LLM workflows, file editing, and agent performance monitoring.

  • Project
    deepagent-ltd/deepagent-code

    DeepAgent Code (DAC) is a desktop‑plus‑terminal workspace that lets you use an LLM as a coding assistant for multi‑step tasks. It offers persistent sessions, inspectable project memory, sub‑agents for independent jobs, a managed browser, and optional instant‑messaging integration, all configurable with any compatible LLM provider. Licensed under AGPL‑3.0.

  • Project
    OthmanAdi/planning-with-files

    A persistent file-based planning system for AI coding agents that stores task state on disk to prevent memory loss during context resets, crashes, or long-running sessions.

  • Project
    MiniMax-AI/minimax-code

    MiniMax Code (CLI `mcode`) is an AI‑driven coding assistant that lets you describe code changes in natural language, then automatically edits, tests, and runs the code in your local project. It works with MiniMax’s own model or any OpenAI/Anthropic‑compatible endpoint, offers interactive TUI, headless scripting, and a desktop UI, and includes a browser‑based demo (pocket‑pet) that shows the full prompt‑to‑working‑app loop.

  • Project
    peterfei/ai-agent-team

    A virtual AI development team with specialized roles (PM, Backend, Frontend, etc.) and a persistent memory system for managing complex software projects.