Agentic Development Environment for Devs Who Don't Ship Slop
Free Local Software · Powered by Claude Code
Website • Features • Why OpenADE • Download • Contributing • Discord
Most AI coding tools throw agents at your codebase and hope for the best. You end up babysitting, fixing hallucinations, and wondering if the "AI-assisted" code is actually worse than writing it yourself.
OpenADE takes a different approach: Plan → Revise → Execute.
By spending time upfront to craft and refine your implementation plan, you can run agents linearly with confidence. No more context-switching. No more "wait, what did it just do?" moments. Just saturated throughput and maximal work done.
We built this as a side project at Bearly AI over a few weeks—within days it was building itself. The ability to drill down on plans makes agent execution feel almost deterministic. It killed todo lists for us—we just fire off initial plans and come back when we're ready. It's now 90% of our screen time, replacing our IDE and terminal.
Describe your task, get a detailed implementation plan, refine it with inline comments, then execute with a single click. Your agents work from a shared understanding—not vibes.
Files, diffs, changes, agent messages—everything is commentable. Point the agent directly at what you want to fix or improve. Type less, iterate faster, ship better.
Information-dense, text-heavy, sharp corners, great syntax colors. If you love the terminal aesthetic, you'll feel right at home.
First-class support for 15+ integrations out of the box: Linear, Notion, Asana, Stripe, Vercel, GitHub, and more. Bring your whole workflow into the loop.
Your code never leaves your machine. No cloud. No syncing to someone else's servers. Works without an internet connection.
Telemetry: Optional anonymous telemetry. Easily disabled in settings.
- Beautiful drafting input with rich text and @ mentions
- Single-key actions for common workflows
- @ file references, skills, and slash commands
Diff viewer, file browser, fast search, full terminal, and process manager—all without leaving the app.
No separate billing. No per-seat licensing. Works with your existing Claude Code subscription.
Coming soon: Codex support
Snappy & Clean. We use the app 8+ hours a day so we make sure it is.
Every execution creates a patch snapshot. Rollback to any point with a click. Never lose work.
Isolate experiments in git worktrees. Keep your main branch clean while agents explore.
Fire off a task, go scroll X and TikTok, get pinged when it's done. We don't judge.
Track your throughput. See what's working. Optimize your flow.
Dark, light, synthwave, or roll your own. Make it yours.
Clean architecture. TypeScript throughout. MobX state management. Built to be extended.
You can't edit code directly in OpenADE.
We know, we know. Here's the thing: after working this way for a while, we've found that agents are fast at making small changes—and they respect your automations, docs, and conventions while doing it. Often they'll catch something you would have missed.
Let them cook. 🧑🍳
| Platform | Download |
|---|---|
| macOS (Universal) | OpenADE-0.52.0-universal.dmg |
| Linux (arm64) | OpenADE-0.52.0-arm64.AppImage |
| Windows | OpenADE-Setup-0.52.0.exe |
We'd love your help making OpenADE better.
git clone git@github.com:bearlyai/OpenADE.git && cd OpenADEcd projects/web && yarn && yarn startcd projects/electron && yarn && yarn startOr, if you're feeling meta: clone it, open it in OpenADE, and start the processes from there. 🔄
Questions? Ideas? Found a bug? Come hang out.
MIT © Bearly AI