An open-source AI coding agent that reads your project, edits files and runs commands and tests from your terminal, desktop, browser or editor.
JetBrains/thinkrail
A desktop and browser IDE built around the pi coding agent, running agent work in isolated git-worktree workspaces with living specs and skills.
Open-source alternative to:CursorWindsurfGitHub Copilot Workspace
- Stars
- 621
- Forks
- 49
- Open issues
- 167
- Last commit
- yesterday
What is it?
ThinkRail is an agentic IDE that wraps the pi coding agent in a workbench with Monaco editor tabs, a live git Changes view, terminals, and documents. Each project can spawn isolated git-worktree workspaces with their own branch and working directory, so agents can work in parallel without touching the main checkout.
The project keeps project knowledge in a graph of SPEC.md files beside the code, which the agent reads and updates through dedicated spec tools. It also loads Agent Skills and bundled workflows for setup, design, and opening pull requests through the user's own gh CLI. MCP servers can be connected from pi's mcp.json or from the settings page.
ThinkRail runs as a native desktop app for macOS, Windows, and Linux, or as a thinkrail CLI that opens the same app in a browser. Users sign in with JetBrains AI or bring their own provider credentials through pi's auth.
Key features
- 01Creates isolated git-worktree workspaces so agents work in parallel without touching the main checkout.
- 02Runs multiple concurrent pi chat sessions per workspace, each with its own model and cost readout.
- 03Maintains a graph of SPEC.md files that the agent reads and updates through spec tools.
- 04Loads portable Agent Skills and bundled workflows for project setup, design, and pull requests.
- 05Connects MCP servers and asks before calls the server does not mark read-only.
- 06Available as a desktop app or a thinkrail CLI that opens the app in a browser.
Use cases
- Running several AI coding tasks on the same repository at once, each in its own branch.
- Keeping product and design intent in versioned spec files that the coding agent can follow.
- Reviewing agent-made changes in a git diff view before merging the branch.
- Using an existing pi setup with its skills and MCP servers inside a dedicated IDE.
Topics
FAQ
- What is thinkrail?
- A desktop and browser IDE built around the pi coding agent, running agent work in isolated git-worktree workspaces with living specs and skills. ThinkRail is an agentic IDE that wraps the pi coding agent in a workbench with Monaco editor tabs, a live git Changes view, terminals, and documents. Each project can spawn isolated git-worktree workspaces with their own branch and working directory, so agents can work in parallel without touching the main checkout.
- Is thinkrail free to use? What license does it have?
- thinkrail is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
- Is thinkrail still maintained?
- Actively maintained. The last commit to thinkrail was on Oct 17, 2026.
- What can I use thinkrail for?
- Running several AI coding tasks on the same repository at once, each in its own branch. Keeping product and design intent in versioned spec files that the coding agent can follow. Reviewing agent-made changes in a git diff view before merging the branch. Using an existing pi setup with its skills and MCP servers inside a dedicated IDE.
- Where can I get thinkrail?
- The source code and installation instructions are on GitHub: https://github.com/JetBrains/thinkrail
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A spec-driven development workflow for AI coding assistants that agrees on proposals, specs, and tasks before any code is written.
An open source MCP server that connects AI agents and IDEs to databases, with prebuilt database tools and a framework for custom tools.
Long-term memory for AI coding agents that lets a session continue across agent CLIs such as Claude Code, Codex and Cursor, backed by a git-tracked markdown wiki.
A collection of Agent Skills and plugins that let coding agents work with Google Stitch design projects via the Stitch MCP server.
An open-source Chinese book on AI Agent design, built around Agent = LLM + context + tools, with runnable companion experiments per chapter.