Roo Code

A whole dev team of AI agents in your editor.

Roo Code Inc. Open source Since

Roo Code (formerly Roo Cline) is an open-source, multi-agent VS Code extension that gives every developer a full AI dev team inside the editor. Fork of Cline with expanded capabilities: multiple specialized agent modes (Code, Architect, Ask, Debug), boomerang orchestration to delegate sub-tasks, MCP support, and support for all major LLM providers. 38K+ GitHub stars as of May 2026.

+ Pros

  • Multi-agent modes (Code, Architect, Ask, Debug) with specialized prompts and tools for each role
  • Boomerang orchestration enables complex multi-step workflows across parallel sub-agents
  • Works inside VS Code, Cursor, Windsurf — no editor lock-in
  • Fully open source Apache 2.0 — 38K+ stars and very active maintenance
  • MCP-native and BYO LLM — complete flexibility on tooling and model provider

Cons

  • Requires LLM API key — no free inference included
  • Setup complexity higher than commercial tools like Cursor or GitHub Copilot
  • Roo Cloud not yet available — cloud execution still requires local setup
  • Multi-agent boomerang workflows can accumulate significant token costs with expensive models

Pricing

Free (OSS)

$0

Open source VS Code extension with basic features. Apache 2.0 license.

Pro

Discontinued

Project shut down May 15, 2026.

Roo Code (originally Roo Cline) is a VS Code extension and fork of Cline that expanded the single-agent model into a full multi-agent dev team inside the editor. Launched in late 2024 and growing to 38K+ GitHub stars by May 2026, it is one of the most actively maintained open-source coding agent projects. The core innovation is agent modes: Code (writes and edits), Architect (designs systems), Ask (answers questions without editing), and Debug (diagnoses issues) — each with its own system prompt, tool access, and context strategy.

Boomerang orchestration lets the active agent spawn and delegate sub-tasks to other agents, each in an isolated context window, and collect the results — enabling complex multi-step workflows without overflowing a single context. Roo Code is provider-agnostic (Claude, GPT, Gemini, local models via Ollama) and MCP-native, allowing it to connect to any external tool server. The extension works inside VS Code, Cursor, Windsurf, and any VS Code fork, and the team is developing Roo Cloud for hosted execution.

Key Features

• Multi-agent modes: Code, Architect, Ask, Debug — each with its own system prompt and tool set
• Boomerang orchestration: spawn sub-agents for isolated sub-tasks, collect results back
• MCP support: connects to any Model Context Protocol server
• BYO LLM: Claude, GPT, Gemini, local models via Ollama
• Works in VS Code, Cursor, Windsurf, and all VS Code forks
• Apache 2.0 licensed — fully open source
• 38K+ GitHub stars (May 2026)
• Active community with frequent releases
• Roo Cloud (coming soon): hosted execution without local setup

Version History

v3.54.0

Final release; repository archived by owner on May 15, 2026. Roo Code extension officially shut down.

v3.53.0

Community handoff announced. Added GPT-5.5 (OpenAI Codex), Claude Opus 4.7 (Vertex AI), checkpoint navigation controls.

v3.52.1

Patch release: added correct JSON schema for .roomodes config files; removed hiring announcement from VS Code extension UI.

v3.52.0

Added Poe as AI provider, xAI Grok-4.20 models, OpenAI GPT-5.4 mini/nano models, MiniMax model listing fixes.

v3.51.1

Patch release: added Cohere Embed v4 for Bedrock, Gemini 3.1 Pro CustomTools on Vertex AI, and GPT-5.4 to ChatGPT Plus/Pro Codex catalog.

Signature Snippet
Open Roo Code in VS Code and switch to Architect mode. Type: 'Design the database schema and API layer for a multi-tenant SaaS billing system.' Architect mode produces a full design doc. Then Roo Code's boomerang orchestration hands off implementation sub-tasks to Code mode agents in parallel — one for the schema migration, one for the API routes, one for the tests — each running in its own context window.

Live feed in your inbox

Track the tools. Lead the shift.

Tech leaders use Artificialus to stay ahead: editorial picks, agent comparisons, MCP updates, and signal-heavy analysis when it matters.

No spam. Only tools and shifts worth tracking.