ThinkRail 0.3.1: bring your own model, your own packages, your own font. Then try to lose a chat

JetBrains-backed ThinkRail released version 0.3.1, adding a command box opened with a double-tap of Shift (or ⌘K/Ctrl+K) that surfaces chats waiting across open projects, plus Settings → Providers → Add server for connecting self-hosted model servers via the OpenAI chat-completions API, with presets for Ollama, LM Studio, vLLM and llama.cpp. The release also lets users install pi packages from Settings → Extensions without a terminal, customize interface and code fonts, and adds Russian to its supported languages alongside English, Spanish, Simplified Chinese and Japanese; code ligatures are now off by default. Fixes include a macOS "Maximum call stack size exceeded" crash mid-reply and duplicate or missing desktop link opens.

ThinkRail https://thinkrail.ai/?utm source=blog&utm medium=organic-social&utm campaign=release-0-3-1 0.3.1 is out. The new release of ThinkRail introduces a few new features that we’ve been working on for some time, as well as a number of fixes and other changes. Here’s what I find the most important among them. One box to find and run things The chat that’s stuck on a question is always in the project you collapsed. Not anymore. Double-tap Shift or ⌘K / Ctrl+K , or the search button in the topbar . An empty box shows what’s waiting for you across every open project, most urgent first. Type a few letters to find any chat, workspace or project, run a command, or jump to a Settings section prov → Providers . It learns which result you pick. This is step one. The plan is to make this box the way you do everything in ThinkRail; files and specs are next. Your own model server, no JSON Company gateway or a model on your laptop: Settings → Providers → Add server . Pick a preset Ollama, LM Studio, vLLM, llama.cpp or paste an address, hit Find models , save. It lands in pi’s models.json , so the pi CLI gets it too. The key never reaches the browser. Limits: OpenAI chat-completions API only, no per-model tuning in the UI. pi packages, no terminal Settings → Extensions → Install package… Paste a pi install line, a pi.dev or npm link, or a name. You see what it adds before you use it. Don’t like it? Switch it off ThinkRail chats only, the pi CLI still loads it , update it or remove it from the same row. Hate our font? Swap it Settings → Appearance → Fonts : interface and code font, weight, size, ligatures. Use installed fonts or drop in your own files. Applies everywhere at once. Heads-up: code ligatures are now off by default. More languages Russian joins English, Spanish, Simplified Chinese and Japanese. Missing yours? Add it under Settings → Language and send it to us. Fixes - macOS: no more “Maximum call stack size exceeded” crash mid-reply. - Desktop links open once, in your browser. Not twice, not never. - No false unread dot when a background job finishes right after your turn. - Resources stops showing finished subagents as Working . - “1.0 MB”, not “1024 KB”. Thanks to everyone who sent PRs, and welcome to first-time contributors @arhipov-om https://github.com/arhipov-om the Russian translation and @kevin9327 https://github.com/kevin9327 . Try it Grab the desktop app below, or see the rest of what ThinkRail does on our main webpage https://thinkrail.ai/?utm source=blog&utm medium=organic-social&utm campaign=release-0-3-1 . The full list of changes is in the v0.3.1 release notes https://github.com/JetBrains/thinkrail/releases/tag/v0.3.1 . Find us on social media platforms: X/Twitter https://x.com/ThinkRailAI , YouTube https://www.youtube.com/@ThinkRailAI , and TikTok https://www.tiktok.com/@thinkrail . You can also browse our GitHub repository https://github.com/JetBrains/thinkrail , take our survey https://forms.gle/es1ksqAax6hnDWCP8 , and join our Discord server https://thinkrail.ai/discord to say “hi ” and let us know what you think. Your feedback is greatly appreciated. Happy developing ThinkRail is backed by JetBrains, leveraging their expertise in building developer tools that are both powerful and intuitive.