stagewise is an open-source, AGPL-3.0 agentic IDE written in TypeScript, developed by stagewise GmbH, that embeds a coding agent directly in the editor and is aimed at developers who want to browse, build, and orchestrate agents in one place while bringing their own model keys.

What it is

stagewise describes itself as "The Open Source Agentic IDE" — an integrated development environment for developers with a coding agent built right in. It sits in the Developer Tools / IDEs & Code Editors category, is written in TypeScript, is developed by stagewise GmbH, and is offered under the AGPL-3.0 licence. It appears in this registry under the topics agent, agent-orchestration, and ide, and its homepage is stagewise.io. The facts record 6,822 stars, 507 forks, 27 open issues, and a most recent push on 2026-09-23.

The concrete problem it addresses is the loop between browsing an application and editing its code. stagewise lets a developer browse and build in the same tool, so there is no context switching, and the coding agent has full access to the active tab's console and debugger. A developer can make temporary test changes or connect a codebase for permanent edits, reverse-engineer any website's components, style systems, and colour palettes, and then view and apply the resulting code changes in a preferred editor through IDE integration. Alongside the editing loop, it replaces per-vendor agent configuration: a single API key connects an existing model subscription, and a custom-provider pathway covers local inference and custom model definitions.

Key capabilities

• Browse and build in the same tool, with no context switching between a browser and the IDE.
• A coding agent with full access to the active tab's console and debugger.
• Temporary test changes, or connecting a codebase for permanent edits.
• Reverse-engineering a website's components, style systems, and colour palettes.
• IDE integration to view and apply code changes in a preferred editor.
• Bring your own API key, fully supported for all AI providers, including custom providers with local inference and custom model definitions.
• Single-key imports for existing subscriptions from Moonshot AI (Kimi), Alibaba DashScope (Qwen Coding Plan), MiniMax, Xiaomi MiMo, and Mistral, plus an OpenRouter key covering 345+ models.

Who uses it and how

• Developers who already pay for a coding subscription such as Kimi, Qwen Coding Plan, MiniMax, Xiaomi MiMo, or Mistral and connect it with one API key instead of opening new billing.
• Teams that need custom or local inference, registering a custom provider and defining models inside stagewise.
• Front-end and product developers who reverse-engineer components, style systems, and colour palettes from existing websites.
• Individuals and small teams using a stagewise Account, which offers a Free tier with limited access to three standard models (Default, Quick, Smart), Pro at $20 per month with access to all models including Frontier and Open-Weights, and Ultra at $200 per month with 15x higher limits than Pro.
• Developers orchestrating multiple agents, which is the workflow implied by the agent-orchestration topic.

Getting started

Download stagewise from stagewise.io (the app is served at ade.stagewise.io) and follow the short onboarding guide to set up an account. After that, either connect an existing subscription with a single API key, bring your own key for a provider such as OpenRouter, or use a stagewise Account tier.

How it compares

The facts for this entry do not name any commercial products that stagewise replaces, nor any similar tools it would sit alongside. On that basis it stands alone in this registry, and the only alternatives listed in the facts are the model providers and account tiers it connects to rather than competing IDEs.

When to use it — and when not to

The documented path is a download from stagewise.io plus onboarding; the facts give no self-hosted server installation, database, storage, or mail configuration for an operator to run, so anyone expecting a documented self-hosting runbook should not assume one exists. Teams that cannot adopt a copyleft AGPL-3.0 licence, or that need build-from-source instructions spelled out in the README, should look elsewhere, because the published README excerpt is short and covers onboarding rather than deployment.