AI Coding Agents: Turning Software Tasks Into Tested Changes

Coding agents turn software requests into a loop of investigation, editing and checking. A model chooses what to inspect or change; tools let it work in the repository; execution results shape the next step. Useful autonomy depends on understanding the task, preserving existing work and checking the exact change that will be integrated. This chapter follows that workflow without treating an agent’s completion message as proof that the software works.

One CLI that takes the working tree you already have, runs it somewhere else (a container, a VM, an SSH host, or a managed sandbox), and streams the result back, so you get an edit-run loop without writing a CI job for every iteration. MIT licensed, coordinator runs on your own Node plus Postgres or on Cloudflare Workers, and it claims 81 providers. Their line is "Crabbox Software Is Free. Compute Isn't."

Herdr is where your coding agents live. However many you run, across however many projects, each in its own terminal. Walk away and they keep working. Come back from any machine and they're where you left them.

Like a tmux thing or something.

M3 reaches frontier-level performance on specialized tasks such as coding and agentic work. It uses MSA (MiniMax Sparse Attention), a new attention architecture proposed by our team, and supports ultra-long context windows of up to 1M tokens. To much anticipation, it is also a natively multimodal model that supports image and video input and can operate a desktop computer.

These three capabilities are now table stakes for closed-source frontier models. M3 is currently the first and only open-weight model to bring all three together.

MiniMax-M3 is a native multimodal model with 1M context. It has ~428B parameters and ~23B activated parameters.

Coding & Cowork Capability: M3 achieves frontier-level performance across long-horizon agentic benchmarks, excelling in both coding and cowork.

The open source AI coding agent

Pi is a minimal agent harness. Adapt Pi to your workflows, not the other way around. Customize Pi with extensions, skills, prompt templates, and themes. Bundle them as Pi packages and share via npm or git.