What problem it solves

Coding agents need isolated places to work without disturbing a checkout, while developers need to steer concurrent sessions and review each change. They also need durable project context that survives beyond a single chat session, along with tools and workflows that fit how they work.

Key features

• Isolated git worktree workspaces — Open a git repository as a project and create workspaces as `git worktree`s with their own branches and working directories. Agents can work in parallel while the original checkout remains untouched.

  Source: README

• Integrated agentic workbench — The workbench combines editable code, project files, terminals, documents, chat sessions, and diffs within the active workspace. Review comments on files and diffs can be sent to chat as structured context.

  Source: README

• Durable project specifications — A connected graph of `SPEC.md` files lives beside the code and is maintained through dedicated tools. The Specs view renders the graph as a tree in the workbench so project intent can remain available across sessions.

  Source: README

• Concurrent steerable coding chats — Each workspace can host several concurrent `pi` sessions with separate models and live token and cost readouts. Developers can steer active turns, queue follow-ups, delegate work to subagents, and edit a shared plan.

  Source: README

• Extensions and integration tools — The project supports portable Agent Skills, bundled workflows, MCP servers, and installable `pi` packages. It also connects to JetBrains AI or other providers and supports OpenAI-compatible custom model servers.

  Source: README

Quick start

Copied verbatim from the project's README.

curl -fsSL https://raw.githubusercontent.com/JetBrains/thinkrail/main/install.sh | bash

thinkrail ~/code/my-repo

git clone https://github.com/JetBrains/thinkrail.git
cd thinkrail
bun install
bun run dev                          # host + web client together; Ctrl+C stops both

Who it may suit

This may suit developers who want to work with coding agents inside an IDE while keeping agent changes isolated, reviewable, and connected to durable project specifications. It may also suit contributors interested in building an agentic desktop or web application around the `pi` coding agent.

Topics to discuss

• How should living specs evolve when multiple agents work on the same project?

• What workflow best balances isolated agent workspaces with direct work in the main checkout?

• Which extension or MCP integrations would most improve day-to-day coding workflows?

Written by ReadmeX AI from the project's README, source and public GitHub data; it may contain mistakes.

Read the original README https://github.com/JetBrains/thinkrail#readme

ReadmeX opened this community for a trending project; it is not run by the project. Maintainers can claim it from the community page.