Let’s Code Aadml IDE
Clone a repo in the browser. The agent reads, edits, and proposes reviewable diffs you accept hunk-by-hunk — then exports a GitHub PR. No silent writes.
Now shipping — Let’s Code Aadml IDE
The Aadml OS for coding & knowledge work
One platform where agents read your code, edit with reviewable diffs, run and verify, retrieve from your knowledge, and automate on a schedule — code-aware, context-aware, and cost-aware. Self-hostable and built for teams.
Edits land as reviewable diffs — accept hunk-by-hunk, never silent writes.
Agent working…
Models from every major lab
OpenAIAnthropicGooglexAIMetaMistral
Everything an agent needs to ship work
A complete, self-hostable platform — not a wrapper. Reason, retrieve, edit, run, and automate in one place.
1,800+ tools
Builtin tools, any MCP server, and any OpenAPI/Postman service — lazily loaded and authorized per agent, so the model never drowns in a giant prompt.
Code intelligence
Symbol- and graph-aware retrieval grounds every change in how your codebase actually fits together — callers, callees, imports, and related tests.
BetaKnowledge & RAG
Ground agents in your uploaded docs and crawled websites with hybrid semantic + keyword search, per agent.
Automation
Fire agents from webhooks (signals) or cron (schedules) with budget caps, retries, dead-letter recovery, and human-in-the-loop approval.
Multi-tenant & cost-metered
Orgs, workspaces, RBAC, encrypted credential vaults, and per-token cost accounting baked in from day one — not bolted on.
From prompt to pull request
The agent decides when to answer, edit, run, or ask — you stay in control.
1
Ask or assign
Start a conversation or trigger an agent from a webhook or schedule. It retrieves the right context first.
2
Edit & verify
The agent proposes minimal diffs against a clone, runs focused checks, and asks when it’s unsure.
3
Review & ship
You review per-file and per-hunk, accept what’s right, and export a pull request — guardrails intact.
1,800+
Tools
6
Model makers
100+
Models
24/7
Autonomous
Ready to deploy your first agent?
Self-host with Docker in minutes, or start on the cloud. Bring your own model keys.