Crush changelog
The last 25 stable releases of Crush, newest first, with the lines that remove, rename or deprecate something pulled out on top. Taken from the GitHub releases every morning.
Breaking changes and deprecations
None of these releases removes, renames or deprecates a flag, command or setting.
Latest releases
-
- completions: OpenAI-compatible APIs
- responses: OpenAI, or anything routed through it
- messages: Anthropic-compatible APIs
- --can-reason is now --thinking always|never|toggleable
- --max-attachments N is new
- Reasoning levels show friendly names in the picker, like "X-High"
- Host themes: the host app can theme the TUI over the server API, so Crush matches your applications.
- Host context: the host can tell Crush what's in view, like open files, and those files appear in completions.
- Append to the next prompt: clients can add text the agent reads with the next prompt, without starting a run or cluttering the chat.
- Shared viewing: clients can see which session other clients are looking at.
-
- Server side improvements
- External editor swallowing empty buffer
- Models behind OAuth couldn't be used if they weren't in Catwalk
- Fixed MCP toggled misbehavior
-
- ChatGPT OAuth now refreshes with Catwalk
- Claude Channels, part 2/3
- Toggle Sidebar on and off with ctrl+b
- Quit dialog improvements and mouse support
- Fixed clipboard false reporting successful copies
- Count Anthropic cache-creating tokens
- Correctly resolving the project data directory in session and stats
-
- Hypercredits are now displayed in real time!
- Added spaces to errors in chat
- Increased default timeout to 2 minutes
- Fixed extended help, while in a mode
- Fixed wrapping while displaying the plan
- Fixed syntax highlighting on long bash calls
- More bang mode and bash tools improvement also by our favorite
-
- fix(clipboard): only report copy success if it happens
- fix: stop long reasoning traces from running away with the heap
- fix(config): honor providers.json when auto-update is disabled
- fix: inline code copying
- fix: flickering screen when inline code was present
- fix(copilot): route grok models through responses api
- fix(discover): keep a user-set supports_attachments in the LM Studio enricher
- fix: assign models from opencode zen
- fix(list): keep AtBottom honest when the top item is taller than the viewport
- fix: make timeout configurable
-
- Qwen3.8-Flash
- Qwen3.8-27B
- Qwen3.8-2.4T-A95B
-
- Fixed ctrl+a after v0.91.0.
- Improved copying by selection with wrapped lines
- Added documentation on how to set a custom Catwalk URL
-
- ctrl+shift+a selects everything
- ctrl+shift+c copies the selection
- ctrl+shift+x cuts the selection
- ctrl+shift+v pastes text
-
- Updated Bubble Tea with a couple fixes for some terminal emulators.
- Updated Fantasy with fixes for AWS SSO and better cache hitting.
- Fixed wrong background color on permissions dialog. This was a regression from adding Bash syntax highlighting on v0.89.0.
- Updated our .gitattributes to make development easier on Windows. We're normalizing LF (instead of CRLF) as line breaks to fix some tests on Windows.
-
- UX: Typing clear in the command will match Summarize; thanks, @joestump!
- LSP: @piakdev fixed a case where LSP could panic
- MCP: Slow loading MCPs could block messages from being sent: we reduce default timeout
- Server-Client: MCP OAuth didn't work in server-client mode; now it does
- Server-Client: Clients in the same workspace could kill each other: not anymore
- Server-Client: When the server is running over TCP keystrokes no longer hit the server and slow everything down
- Hyper: Added various small improvements for better UX
-
- lsp_rename - Rename a symbol and all its references in the project
- lsp_replace_symbol - Replace, insert before/after, or delete an entire function/method/class by name
- When focused on the chat view: l or right arrow focuses on the sidebar
- j/k or up/down arrow scrolls up and down
- g scrolls to top shift+g scrolls to the bottom
- h or left arrow focuses back on the chat view
- tab always returns focus to the editor
- Reasoning effort level none is now working for Baseten models
- Thinking blocks are now rendering for MiniMax M3 via OpenCode (thanks @marifcelik)
-
- Fixed: crush auth now works outside of client/server mode
- New: Crush now auto retries auth under the hood when it fails
- Improved: dialogues now render great on tiny screens
- Perf: scrollbar optimizations for faster rendering
- Perf: memoized markdown renderer for faster rendering
- Perf: short circuit jq builtin operations if the turn was canceled before it started
- Yay! OpenAI merged our PR upstream so we're back off our fork
-
- https://github.com/charmbracelet/crush/pull/3282/changes/5a7899eeba613ca0ddf3c03fdcf227cec0f85f36 chore(deps): bump fantasy (0.35.2 -> 0.36.0) (@coloryourlife)
Other agents
Claude CodeCodex CLIGemini CLICopilot CLIopencodeHermes AgentQwen CodeKimi CLIGooseMistral VibeAiderCline All agents →
Run Crush next to the other agents and see when one needs you: AgentVera keeps them in one window on Mac, Windows and Linux, and on your phone.
Bring your agents to one desk.
Download, pick a project, open your first agent. Your installed CLIs are ready to go.