> For the complete documentation index, see [llms.txt](https://docs.perception.cx/perception/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.perception.cx/perception/enma-ai/overview.md).

# Overview

Enma is a desktop multi-agent AI coding app for Windows. You bring your own models, local or cloud. Point it at a project folder and a team of agents plans, codes, tests, reviews and generates assets for you.

<figure><img src="/files/TKc0uTSrVD921ARSw92C" alt=""><figcaption></figcaption></figure>

## Why Enma

* **Token efficient.** Prompt caching, targeted file reads, context advisories and automatic compaction keep long sessions cheap.
* **Multi-agent.** A planner delegates to specialist agents that work in parallel. You can message any of them mid-run.
* **Your models.** Any OpenAI-compatible endpoint works: LM Studio, Ollama, llama.cpp, OpenRouter, or your own gateway. Run each agent on a different model.
* **Safe by default.** Ask-first permissions, per-turn checkpoints, isolated runs on a throwaway copy of your repo, and a git safety floor. Agents never push.
* **Beyond code.** Images, sprites, 3D models, skyboxes, audio and video straight into your project, plus a vision agent that can playtest your game.

## Download

[Download the installer](https://perception.cx/percapi/enma/harness/pub/enma_harness.exe)

Windows 10/11, per-user install, no admin needed. Installed copies update themselves in the background.

Start with [Getting Started](broken://pages/ed6270139f12e638d8cd74a1f3c2c679e670049c).