Wails + Go v0.4.0 Desktop + Android

chidori

AI-powered IDE with a built-in reasoning engine.

Ask, Agent, Plan, or Debug — local and remote LLMs, file explorer, editor, and terminal in one app. Pair an Android phone as companion or inference node.

See it in action ↓

Download v0.4.0 Getting started GitHub Android APK
chidori ~/projects
chidori Ask mode — code editor with inline AI chat

Ask mode — editor + AI chat side by side

4 AI ModesAsk, Agent, Plan, Debug
Hybrid LLMLocal + remote inference
File ExplorerProject management
Built-in TerminalIntegrated shell
Desktop application. chidori is a native macOS/Windows IDE that runs on your machine. Built with Wails, it provides a modern code editor with AI reasoning capabilities integrated directly into your development workflow.

Screenshots

Real screenshots, not mockups (captured from a live coordinator session). Click any image to enlarge.

Ask mode — code editor with inline AI chat Click to enlarge
01 — Ask Ask mode chidori has the file tree, editor, and chat open together. You ask a question; the answer streams next to the code it just read.
Planner settings showing live coordinator topology Click to enlarge
02 — Route Planner topology The coordinator shows which inference nodes are attached and where the next request will go — local Ollama or a remote worker.
Inference Source settings for attaching LLM nodes Click to enlarge
03 — Attach Inference Source Point at Ollama, LM Studio, or a machine on your LAN by IP. No cloud account required.
Model picker listing cloud and local models Click to enlarge
04 — Pick Model picker Choose a hosted or local model per node. Same Ask / Agent / Plan / Debug flow either way.

Download

Pick a platform. Desktop packages ship from xdutsuay/chidori (releases only). The Android companion APK ships from xdutsuay/chidori-nagasa.

macOS

Apple Silicon

M1–M4. Unzip and open chidori.app. Not code-signed yet — right-click → Open on first launch.

v0.4.0 | arm64 | .zip

Download for Mac
curl -LO https://github.com/xdutsuay/chidori/releases/download/v0.4.0/chidori-macos-arm64-v0.4.0.zip

Windows

x64 desktop

Unzip and run chidori.exe. No installer — portable folder.

v0.4.0 | amd64 | .zip

Download for Windows
curl.exe -LO https://github.com/xdutsuay/chidori/releases/download/v0.4.0/chidori-windows-amd64-v0.4.0.zip

Android

Companion APK

On-device GGUF chat, LAN pairing, coordinator monitor, remote chat, optional phone-as-node. Debug-signed sideload.

nagasa 2.1.2 | arm64 | .apk

Download APK

System requirements

  • macOS — 10.10+, Apple Silicon only for the published zip
  • Windows — 64-bit Windows 7+
  • Android — phone or tablet that can sideload APKs; enable Install unknown apps for your browser/file manager
  • Optional — Ollama or another local OpenAI-compatible server for desktop inference

All desktop releases · All Android releases · Android source (public)

macOS may warn that chidori is not code-signed. Here's how to open it.

Right-click chidori.app → Open → confirm Open. Gatekeeper does this on first launch for unsigned apps — it is not a malware scan result. Signing and notarization are not in v0.4.0.

Get updates

No mailing list yet. Watch GitHub releases or add the Atom feed to any reader (many readers can email you). Same feed as the changelog.

Watch releases on GitHub RSS / Atom Changelog

What's new in v0.4.0

High-reliability desktop checkpoint on the public chidori GitHub repo (macOS arm64 + Windows x64), plus the Android companion APK. Full notes on the changelog.

  • Stop actually stops Shared Ask/Agent/Plan/Debug deadlines (TTFT, stream-idle, total-run) and packaged Stop cancel — runs no longer hang after you hit Stop.
  • Oversize prompts fail loudly Fail-closed context budget: too-large prompts return HTTP 400 unless you opt into truncate, instead of silently dropping code.
  • Faster re-index after edits Incremental workspace RAG skips unchanged files, replaces edits, and drops deletes.
  • Broken tools error out Grok ACP rejects terminal RPCs instead of stub-succeeding; MCP is fail-closed. Zen / secondary sidebar landed too.
  • Android companion APK Pair over LAN, monitor runs, remote chat, optional phone-as-node.

What's next

Not a dated v0.5 promise — these are the next public pieces already in motion.

  • Shipped
    Android companion APK Download from Companion mobile app or the Download grid — LAN pairing, monitor, remote chat, optional phone-as-node.
  • Next
    CLI inside the packaged app chidori ask on PATH from the .app, without a separate install step.
  • Later
    Signed macOS builds Notarization so Gatekeeper stops asking you to right-click Open. No date yet.

AI Reasoning Modes

Four integrated modes for different coding tasks — each with direct access to your codebase and local/remote LLMs.

IDE Features
  • Code EditorFile tree, syntax highlighting
  • TerminalIntegrated shell access
  • SettingsLLM config, inference source
  • Project MgmtWorkspace indexing
  • UI ThemeDark mode native
AI Modes
  • AskDirect Q&A with codebase
  • AgentAutonomous coding tasks
  • PlanArchitecture & investigation
  • DebugError diagnosis & fixes
  • InferenceLocal or remote LLMs

Getting started

Prerequisites: macOS 10.10+ or Windows 7+. Optional: Ollama for local LLM inference.

  1. Download

    Get the Apple Silicon zip from the Download section (or latest release).

  2. Extract

    cd ~/Downloads
unzip chidori-macos-arm64-v0.4.0.zip

  3. Launch

    Open Finder and double-click chidori.app. If macOS blocks it, use the Gatekeeper steps in Download.

  4. Configure LLM (optional)

    Go to Settings → Inference Source to connect to Ollama or a remote LLM API.

LLM Configuration

chidori supports multiple inference sources:

Local Ollama

http://localhost:11434

Run Ollama locally and chidori will auto-detect it.

Remote API

https://api.openai.com/v1

Configure any OpenAI-compatible API endpoint.

Custom Endpoint

Self-hosted LLM

Point to your own LLM server running anywhere.

Hybrid Mode

Local + Remote

Use local LLM primarily, fallback to remote.

Architecture

Show the component stack 
┌─────────────────────────────────────┐
│         chidori IDE                 │
├─────────────────────────────────────┤
│ File Explorer │ Editor │ Terminal   │
├──────────────────────────────────────┤
│     AI Reasoning Engine              │
│  (Ask / Agent / Plan / Debug)        │
├──────────────────────────────────────┤
│   Inference Coordinator              │
│  (Local Ollama / Remote API)         │
├──────────────────────────────────────┤
│ LLM Backend (Hybrid Local+Remote)    │
└─────────────────────────────────────┘

Key Features

File Tree

Project Explorer

Browse and open files in your workspace

Code Editor

Syntax highlighting

Familiar code editing experience

Terminal

Integrated Shell

Run commands without leaving the app

Settings

Configuration

Manage LLM, workspace, and editor settings

How it works — components

Native Desktop App

Built with Wails — Go backend, vanilla JS frontend. Native macOS and Windows binaries.

AI Modes

Four interaction patterns — Ask for answers, Agent for autonomous tasks, Plan for architecture, Debug for error resolution.

Inference Coordination

Unified interface for local (Ollama) or remote (OpenAI, custom) LLM endpoints. Auto-detection and fallback support.

IDE Features

Full editor with file tree, syntax highlighting, integrated terminal, and configuration management built-in.

Companion mobile app

Android APK is available now. Install chidori-nagasa, open Settings → Chidori Desktop on the phone, and pair with Companion on the desktop (port 8027). You can monitor runs, chat through desktop models, and optionally offer the phone as an inference node. Source is public on GitHub; desktop IDE source stays private — binaries only via xdutsuay/chidori.

Android

Get the APK

Debug-signed build from GitHub Releases. Sideload — no Play Store account required.

latest | .apk

Download APK

Source

chidori-nagasa

Fork of LM Playground with LAN pairing, monitor, remote chat, and phone-as-node. MIT licensed.

github.com/xdutsuay/chidori-nagasa

View source

Pairing

Same Wi‑Fi

Desktop Settings → Companion must be listening. Phone discovers via mDNS or you type host:8027.

protocol | port 8027

Desktop first

Getting started · Configuration · Changelog · LLM Lab (legacy)

← Home