github.devotd.
NewCoding agents

xiaomimimo/mimo-code v0.1.15 — Terminal Coding Agent and Persistent Memory

The latest release v0.1.15 of xiaomimimo/mimo-code, a terminal coding agent published under the MIT license, and its persistent memory feature.

Repositoryxiaomimimo/mimo-code ↗

The latest release v0.1.15 of xiaomimimo/mimo-code, a terminal coding agent published under the MIT license, and its persistent memory feature.

GitHub description: MiMo Code: Where Models and Agents Co-Evolve

stars
★ 13.7K
Forks
1.5K
Language
TypeScript
License
MIT
Latest release
v0.1.15
Read original source ↗View on GitHub ↗

xiaomimimo/mimo-code is a terminal coding agent built on the MiMo model. The repository name and description put forward the idea that the model and the agent evolve together. It is a public implementation that reads and writes code, runs commands, performs tool operations, and even manages Git inside the terminal, and the latest release is v0.1.15. However, no individual change entries remain from this release, so what actually changed has to be checked directly in the repository.

A terminal coding agent is a tool that finishes code editing and command execution inside the terminal. One of the tool's main features is a persistent memory system that keeps project context even after a session ends. It carries the context about the project over to the next task even when the session is broken, meaning the same repository can be worked on continuously across different sessions. The README explains that this memory makes it possible to keep working on the same project across multiple sessions. Although it is based on MiMo, it can also connect to the APIs of major language model providers, so the model to use can be chosen from a wide range. The implementation language is TypeScript and the license is MIT.

However, the evidence is thin. The version is at the 0.1.x stage, and the recorded score on a 10-point scale is a low momentum 2.4, with the freshness, technical, and quality scores left unfilled. The repository's star count of 13693 and creation date of 2026-06-10 alone tell nothing about the performance of a coding agent. No measurements or benchmarks to gauge performance are given either. For now, treat it as an early-stage tool worth examining the structure of, and it is better to check the release notes before using it for real work.