GitHub - nandandas2407-web/orin-ide: A fully-featured AI-powered IDE that runs anywhere — your phone, tablet, or laptop. Just code.

A full AI-augmented code editor that runs in your browser — terminal, file system, and multi-model AI, all local.

Docs · npm Package · Report a Bug · Request a Feature

What is OrinIDE?

OrinIDE is a fully-featured, AI-augmented code editor that runs entirely in your browser — backed by a lightweight local Node.js server. There's no cloud account, no IDE subscription, and no upload of your code anywhere.

Run one command:

npx orin-ide

Open localhost:3000, and you have a complete coding environment — real terminal, full file system, multi-model AI chat, multi-agent AI collaboration, a diff viewer, and a lot more.

Built by Nandan Das — runs on desktop, laptop, and even Android via Termux.

Preview

_{🎥 Demo video not attached yet — drop preview.mp4 into assets/ and it will show up here automatically.}

Quick Start

# Zero install — run instantly
npx orin-ide

# Or install globally
npm install -g orin-ide
orin-ide

# Custom port
orin-ide --port 8080

Then open http://127.0.0.1:3000 in your browser.

Android / Termux Setup

Run OrinIDE directly on your Android phone using Termux.

# 0. Enable storage access (IMPORTANT)
termux-setup-storage

# 1. Update packages + install Node.js
pkg update -y && pkg install nodejs-lts -y

# 2. Install OrinIDE
npm install -g orin-ide

# 3. (Optional) Install Python & C/C++ compiler support
bash $(npm root -g)/orin-ide/setup.sh

# 4. Start OrinIDE
orin-ide

Then open http://127.0.0.1:3000 in your mobile browser.

Features

🤖 AI Assistant


Multi-Model AI Chat	Chat with AI models via OpenRouter — DeepSeek, Gemini, GPT, Claude, Gemma, and more.
Streaming Responses	Real-time token-by-token streaming with auto-scroll and scroll-lock detection.
Chat Modes	Chat, Generate, Edit, Patch, and Explain — each mode tailors the AI's output format.
Patch Mode	AI outputs surgical `@@patch` diffs instead of full file rewrites.
AI Continuation	Auto-resumes from exactly where it left off if a response gets cut off — verified against the model's actual finish signal, not a guess, so it never silently hands you half a file.
Project-Wide Context	AI reads every file in your project before each message.
System Prompt Engineering	Format rules placed first for free-model compliance.
Model Picker	Switch models from the top bar. Shows token usage, context limits, and model health.
Chat Sessions	Multi-session system with tags, pin/unpin, rename, export/import, stats, and search.
@Skill Mention	Type `@` in chat for an autocomplete dropdown of all skills.

🕸️ Agentic Mode — Multi-Agent Collaboration

A team of four specialized AI agents work your request end-to-end, in sequence, each building on the last:

Architect  →  Coder  →  Reviewer  →  Integrator
 (plans)      (builds)   (audits)     (ships)


Autonomous Task Execution	Hands a full multi-step build to the agent team — planning through implementation — with no micromanagement.
Per-Agent Model Selection	Assign a different model to each of the 4 roles from the config panel.
Multi-Model Collaboration	Each agent sees the previous agents' full output as context, budgeted to fit the model's window.
Self-Healing Generation	If any agent's step is cut off mid-file, it auto-continues from exactly where it stopped before handing off — so a truncated response never gets silently passed down the pipeline or applied to your project.
Terminal-Aware	Agents can emit `@@cmd:` shell commands (`npm install`, `git init`, etc.) — run individually or all at once, streamed live from a real shell.
Session Memory	Every collaboration is auto-saved as a session — rename, switch, or delete from the sessions panel.

🧠 AI Providers & Local Models


Multi-Provider Support	OpenRouter, Anthropic, OpenAI, Groq, Together AI, Ollama (local), DeepSeek — all configurable.
Bring Your Own Provider (BYOP)	Add any OpenAI-compatible API endpoint.
Local Model Support	Run models locally via Ollama — TinyLlama, Mini models, and other lightweight LLMs.
Seamless Model Switching	Switch between local and cloud models on the fly.
Custom API Endpoints	Point to any OpenAI-compatible server — local or remote.
Provider Health	Monitor provider status and model availability from the settings panel.

🐞 AI Bug Fixer


Auto-Detect Errors	AI-powered bug detection scans your code for issues.
Explain Issues	Plain-language explanations of what went wrong and why.
Suggest Fixes	Generates corrected code with explanations for each fix.
Generate Corrected Code	Produces ready-to-apply corrected code with full context.

✨ AI Inline Features


AI Explain on Hover	Ctrl+Hover any word for a rich formatted tooltip with explanation.
Lightbulb Actions	AI-powered code actions in the editor gutter — Fix, Explain, Refactor, Optimize.
AI Code Review	Full-file review with inline severity-tagged comments.
AI Refactor	Select code, describe the change, get a diff-previewed refactor.
Ghost Text Completions	Inline AI completions as you type, Tab to accept.
AI Commit Messages	Generates a conventional commit message from your staged diff.
Natural Language Terminal	Describe what you want in plain English, get the shell command.

📁 File Management


File Explorer	Complete project file tree — create, rename, move, delete files and folders.
File Favorites	Pin/unpin files to a Favorites section at the top of the file tree.
Global Search	Project-wide keyword search with case-sensitive, whole-word, and regex options.
Find & Replace	Cross-file find and replace with highlighted matches.
ZIP Export / Import	Export any project as a `.zip`, or import one instantly.
Media Upload	Drag-and-drop images, videos, and audio into your project.
Quick Open	`Ctrl+P` opens a fuzzy file search modal.

💻 Terminal


Real Terminal	Fully interactive xterm.js shell — not a fake console.
Multiple Sessions	Persistent terminal sessions that survive page refresh.
Session Bar	Switch between terminal sessions with a session bar.
Terminal Themes	Terminal colors sync to match the editor theme.

🌿 Git Integration


Source Control Panel	Shows branch name, staged/unstaged/untracked files.
Stage / Unstage	Stage or unstage individual files or all at once.
Commit	Commit with message (`Ctrl+Enter`).
Push / Pull	Push and pull with status feedback.
Git Init	Initialize a new repository.
Auto-Refresh	Git sidebar auto-refreshes when it becomes visible.

🎨 UI & Experience


Dark Purple Theme	Deep-purple, cyberpunk-inspired UI.
Drag-to-Resize Panels	Resize sidebar, chat panel, agents panel, and terminal by dragging.
Sidebar Positioning	Dock the Explorer and AI Chat to the left or right independently.
Command Palette	`Ctrl+Shift+P` opens a searchable command palette.
Voice Commands	Web Speech API for voice-driven IDE actions.
Live Preview	Opens an iframe-based live preview of the project's `index.html`.
Diff Viewer	Review AI edits line-by-line in split or unified view.
Responsive Layout	Fully responsive — works on desktop, tablet, and mobile.
Mobile Navigation	Tab bar, FAB menu, and touch-optimized controls for Android.

Keyboard Shortcuts

General	Editor	AI Features	Navigation	Terminal
`Ctrl+P` Quick Open	`Ctrl+D` Select next occurrence	`Ctrl+Shift+I` Toggle AI completions	`Ctrl+Shift+F` Search sidebar	Ctrl+` Toggle terminal
`Ctrl+Shift+P` Command Palette	`Ctrl+/` Toggle comment	`Ctrl+Shift+R` Toggle AI code review	`Ctrl+Shift+G` Git sidebar	`Ctrl+Shift+C` Copy selection
`Ctrl+S` Save file	`Shift+Alt+F` Format document	`Ctrl+Shift+E` AI refactor selection	`Ctrl+Shift+S` Snippet Palette	`Ctrl+Shift+V` Paste
`Ctrl+W` Close tab	`Ctrl+\` Split editor	`Ctrl+Hover` AI explain on hover	`Ctrl+Shift+H` Find & Replace
`Ctrl+N` New file	`Ctrl+G` Go to line	`Ctrl+L` Toggle AI Chat panel
`Ctrl+B` Toggle sidebar	`F12` Go to definition	`Ctrl+Shift+A` Toggle AI Agents panel
`Ctrl+,` Open Settings
`Ctrl+K` Shortcuts Panel

AI Models

All models below are available on the OpenRouter free tier unless noted:

Model	Type	Notes
`openrouter/auto`	Auto	Auto-selects best free model (default)
Poolside Laguna xs.2	Free	Fast coding-focused model
GLM-4.5 Air	Free	Z-AI free model
Tencent HY3	Free	Preview model
GPT-OSS 120B	Free	OpenAI open-weights model
Nvidia Nemotron 120B	Free	Powerful free model
Google Gemma 3 27B	Free	Google's open model
Cohere North Mini Code	Free	Free coding model
DeepSeek V4 Pro	Paid	Best overall for code
GPT-5.5	Paid	OpenAI flagship
Claude Opus 4.6	Paid	Anthropic flagship

You can also enter any custom OpenRouter model ID directly in the settings panel.

Project Structure

orin-ide/
├── backend/
│   ├── server.js                  # Express server + WebSocket + AI proxy
│   ├── routes/
│   │   ├── files.js               # File system API + surgical patch endpoint
│   │   ├── terminal.js            # PTY terminal over WebSocket
│   │   ├── export.js              # ZIP export, Termux export
│   │   └── preview.js             # Live preview route
│   └── services/
│       ├── terminalManager.js     # PTY session management
│       └── watcherManager.js      # File change watcher (chokidar)
├── frontend/
│   └── public/
│       ├── index.html             # Main app shell + splash screen
│       ├── assets/
│       │   ├── logo-mark.svg      # Full OrinIDE logo
│       │   ├── logo-icon.svg      # 24px compact logo
│       │   └── favicon.svg        # Browser favicon
│       ├── css/                   # 9 modular stylesheets
│       │   ├── main.css           # Core layout, variables, splash
│       │   ├── sidebar.css        # Explorer, file tree, git panel
│       │   ├── editor.css         # Editor, tabs, diff viewer
│       │   ├── terminal.css       # Terminal panel
│       │   ├── chat.css           # AI chat panel
│       │   ├── modals.css         # All modal dialogs
│       │   ├── mobile.css         # Mobile responsive
│       │   ├── agents.css         # AI Agents panel
│       │   └── overrides.css      # Fine-tuning overrides
│       └── js/                    # 44 feature modules
│           ├── app.js             # Main bootstrap + boot sequence
│           ├── api.js             # API client + AI proxy
│           ├── editor.js          # Monaco Editor manager
│           ├── chat.js            # AI Chat with streaming
│           ├── utils.js           # Utilities, parseFiles, parsePatches, renderMD
│           ├── theme.js           # 4 editor themes with token rules
│           ├── aiexplain.js       # AI explain on hover
│           ├── ainline.js         # Lightbulb, long-press, context menu
│           ├── aireview.js        # AI code review
│           ├── airefactor.js      # AI refactor inline
│           ├── aicodeactions.js   # AI code actions provider
│           ├── agents.js          # Multi-agent collaboration
│           ├── aiterminal.js      # Natural language to shell
│           ├── aicommit.js        # AI commit messages
│           ├── aicompletions.js   # Ghost text completions
│           ├── filetree.js        # File explorer
│           ├── fileicons.js       # 40+ file type icons
│           ├── terminal.js        # xterm.js terminal
│           ├── termthemes.js      # Terminal color themes
│           ├── voicecommands.js   # Voice commands
│           ├── voice.js           # Voice input for chat
│           ├── skills.js          # Skills system
│           ├── features.js        # Chat sessions manager
│           ├── snippet*.js        # Snippet engine + palette
│           ├── shortcuts.js       # Shortcuts panel
│           ├── favorites.js       # File favorites
│           ├── breadcrumb.js      # Breadcrumb navigation
│           ├── spliteditor.js     # Split editor
│           ├── gitdiff.js         # Git diff decorations
│           ├── git-mgr.js         # Git source control panel
│           ├── cmdpalette.js      # Command palette
│           ├── quickopen.js       # Quick open file search
│           ├── providers.js       # AI provider configs
│           ├── preview.js         # Live preview
│           ├── media.js           # Asset upload
│           ├── imageapi.js        # Free image API
│           ├── resizer.js         # Panel drag-to-resize
│           ├── global-search.js   # Global search
│           ├── sidebar-position.js # Sidebar docking
│           ├── autosave-indicator.js # Dirty state indicator
│           ├── vibe.js            # Diff viewer, code snapshots, startup
│           └── orin_patches.js    # Final patches + provider setup
├── bin/
│   ├── orin-ide.js                # CLI entry point
│   └── postinstall.js             # Post-install message
├── assets/
│   ├── logo-mark.svg              # Full logo (root copy)
│   ├── preview.png                # Project preview screenshot
│   └── preview.mp4                # Demo video (add your own)
├── setup.sh                       # Termux dependency installer
└── package.json

Security

Version 1.0.5+ includes a full security hardening pass:

Path traversal protection on all export routes via safeProjectName() validation
Content-Disposition header injection fixed
targetDir restricted to an allowlist of safe roots on Termux export
localOnly middleware added to all routes
WebSocket connection limit (max 10 simultaneous clients)
Broadened rm ban pattern to catch all dangerous variants
multer upgraded to ^2.0.0 — resolves 7 high-severity CVEs from 1.x
archiver upgraded to ^7.0.1 — removes deprecated transitive dependencies

Requirements


Node.js	`>= 18.0.0`
API Key	OpenRouter (free tier available) for AI features
Browser	Any modern browser — Chrome, Firefox, Edge, Brave

Contributing

Contributions, issues, and feature requests are welcome.

Fork the repo
Create a feature branch — git checkout -b feat/your-feature
Commit your changes — git commit -m "feat: add your feature"
Push to the branch — git push origin feat/your-feature
Open a Pull Request

Please check the issues page before opening a new one.

License

See LICENSE for full terms.

Changelog

v1.0.9 — Agentic Mode Crash Fix

Agentic Mode

Fixed a bug where every agent turn threw ReferenceError: systemContent is not defined, which made Agentic Mode fail immediately on every send (the multi-agent pipeline never actually produced output)
The agent's system prompt is now correctly preserved when the conversation is trimmed to fit the model's context budget, instead of being rebuilt from a variable that didn't exist in that scope
Agentic Mode (Architect → Coder → Reviewer → Integrator) is now fully functional end-to-end, including auto-continue on cut-off responses and running @@cmd shell commands from the results

v1.0.8 — Local AI Models, Agentic Mode, Bug Fixer, UI Redesign

Local AI Model Support

Integrated local model execution directly inside Orin IDE via Ollama
Support for running Ollama models, TinyLlama, Mini models, and other lightweight local LLMs
Install, manage, and run local models directly from the IDE without leaving the workspace

Agentic Mode

New Agentic Mode for autonomous and multi-step task execution
Multiple mini AI models work together for improved reasoning and task completion
Intelligent orchestration between models for coding, debugging, and assistance
Phase-based workflow: Thinking → Planning → Coding → Integrating → Done
Fixed a crash that could stop the agent pipeline after the first step
Agents now auto-recover from a cut-off response before handing off to the next agent

AI Bug Fixer

AI-powered bug detection that scans code for issues
Suggests fixes with explanations for each detected problem
Generates corrected code ready to apply

Extensible AI Ecosystem

Bring-your-own-provider (BYOP) support — use any OpenAI-compatible API
Custom API endpoints and self-hosted AI services fully supported
Flexible configuration system for developers and power users
Seamless switching between local and remote models

UI & UX Enhancements

Complete UI redesign and visual improvements
Enhanced animations, layouts, and micro-interactions
Improved accessibility across the entire IDE
Better responsiveness and overall user experience

Accessibility & Integration

Unified AI experience across editor, terminal, and assistant panels
AI features accessible throughout the IDE
Keyboard-first design — every feature accessible via shortcuts

Details

Patch Mode

Patch button in top bar and extras popup — forces AI to output only @@patch diffs
Dedicated patch-only system prompt with zero file creation instructions
Full file content sent in user message for maximum AI context
4-strategy auto-converter catches AI output in any format
LCS-based diff engine finds ALL changed regions
Backend fuzzy matching for resilient patch application

AI Continuation System

Auto-chains when generation is incomplete, verified against the model's real finish reason
Up to 10 continuations with minimal, bounded context per continue
Stop button kills the continuation loop
Fallback "Save to Project" panel when continuation fails

System Prompt Improvements

Format rules placed first for free model compliance
Shorter, more direct prompts
Free model context limits corrected to 16384 tokens

Bug Fixes

DiffViewer.captureOrig() crash fixed
False-positive Continue button eliminated
Patch apply backend now handles CRLF, tabs, trailing whitespace via fuzzy matching
Removed a duplicate continuation handler that ignored the continue limit and could exhaust the model's context window on long sessions

v1.0.7 — Project-Wide AI Context, Surgical Edits, Skills System & More

New Features

Project-Wide AI Context — AI reads every file in your project before each message
Surgical AI Edits — @@patch: format for targeted changes without full rewrites
Skills System — Expert AI behavior presets with @mention autocomplete
HTML Preview in Chat — Live preview button on every HTML code block
Free Image API — 20+ categories from Picsum Photos and LoremFlickr
Image Picker — Visual grid browser for stock images
Smart Portfolio Generation — Real image URLs injected into AI prompts

Bug Fixes

Diff viewer blank screen fixed
Code block header layout fixed
Streaming render fix with requestAnimationFrame batching
Auto-scroll pause on manual scroll during streaming

v1.0.6 — Security Hardening

Full security audit and hardening release. See the Security section for details.

Built with care by Nandan Das

Name		Name	Last commit message	Last commit date
assets		assets
backend		backend
bin		bin
frontend/public		frontend/public
node_modules		node_modules
DESIGN.md		DESIGN.md
DEVLOG.md		DEVLOG.md
LICENSE		LICENSE
README.md		README.md
package-lock.json		package-lock.json
package.json		package.json
setup.sh		setup.sh
start.sh		start.sh

Navigation Menu

Repository files navigation

Table of Contents

What is OrinIDE?

Preview

Quick Start

Android / Termux Setup

Features

Keyboard Shortcuts

AI Models

Project Structure

Security

Requirements

Contributing

License

Changelog

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

Table of Contents

What is OrinIDE?

Preview

Quick Start

Android / Termux Setup

Features

Keyboard Shortcuts

AI Models

Project Structure

Security

Requirements

Contributing

License

Changelog

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages