v0.4.10

Changed

• 📁 File browser drag-and-drop completely reworked. You can now freely drag and drop files and folders anywhere — into folders, onto breadcrumbs, or between nested directories. Multi-select drag, auto-expand on hover, and full expanded-area highlighting are all supported.

Fixed

• 📋 Dropdown menus no longer get cut off by the keyboard on mobile. Menus that open upward (like the model picker) now anchor from the bottom of the screen instead of the top, so they stay fully visible even when the on-screen keyboard is open.
• 🔄 Renamed files no longer break git operations. Renaming or moving a file could cause errors when staging, discarding, or viewing changes — the file list would show garbled entries like "R100 old-name.txt" instead of the actual file. This is now handled correctly.

v0.4.9

Fixed

• 📱 Mobile keyboard no longer covers the chat. On iOS and mobile browsers, opening the on-screen keyboard could push content behind it or clip the bottom of the screen. The layout now properly adjusts to keep everything visible above the keyboard.
• 🗑️ Discarding staged files works correctly. Previously, discarding a file that was already staged (added to a commit) could leave it in a broken state. Staged changes are now properly unstaged and cleaned up in one step.
• 🔄 Git file list no longer flickers when staging. Toggling a file between staged and unstaged could briefly cause a visual glitch where the list jumped. The file list now updates smoothly.

Improved

• 📋 Dropdown menus stay on screen. Dropdown menus (like the model picker) now correctly reposition themselves to stay within the visible area, especially when the keyboard is open or on smaller screens.
• 📖 README updated with badges and clearer tunnel docs. Added project badges (stars, language, Discord) to the top of the README, and clarified the remote access / tunnel instructions.

v0.4.8

Fixed

• 💬 Clicking chats now opens them instantly. Tapping a chat in the sidebar, clicking "View Chat" on an automation run, or following a notification toast now takes you straight to the conversation. Previously these actions could sometimes get stuck or fail to switch to the right chat.

v0.4.7

Fixed

• 🛡️ Chat no longer breaks after crashes or interruptions. If a previous session ended unexpectedly, leftover data could cause permanent errors when resuming a conversation. The chat now automatically cleans up mismatched data on load so you can always pick up where you left off.
• 🔄 Better compatibility with OpenAI models. Fixed issues where certain internal data was accidentally sent to OpenAI, causing requests to fail. Conversations with tool use now work reliably across all supported providers.
• 🪵 Clearer error reporting for AI requests. When an AI request fails, the error details are now logged properly instead of being silently swallowed, making issues easier to diagnose.

v0.4.6

Added

• 🎯 Default model. Pick a default model from Settings › Models. New chats and gateway requests will use it automatically instead of whichever model happens to be first in the list.

Changed

• 🔄 Smarter gateway model selection. The gateway now respects your default model setting before falling back, and can auto-discover available models from providers that don't have a pre-configured list.

v0.4.5

Fixed

• 🧠 Reasoning models no longer lose chain of thought across parallel tool calls. Multiple tool calls in the same turn are now grouped into a single assistant message with shared reasoning items, instead of being split into separate messages that broke reasoning model round-tripping (e.g. o3, o4-mini).
• 🛡️ Orphaned tool calls no longer break chat history. If a crash or data corruption left a function call without a matching result, the Responses API conversion now skips the orphaned call instead of sending invalid input that caused permanent 400 errors.
• 🔄 Tool output ordering prevents corrupted history on crash. Tool call outputs are now appended to the output list before marking the call as "completed", so a crash between the two steps can no longer produce a "completed" call with no output — which would corrupt the message history on reload.
• 🧹 In-progress tool calls scrubbed on error or cancellation. When a chat task is cancelled or hits an error, any tool calls still marked "in_progress" are now set to "failed" before persisting, preventing stale in-progress items from lingering in the database.

v0.4.4

Added

• 🧙 Setup wizard. A friendly first-run guide walks you through picking a folder and connecting your AI. Pops up automatically after sign-up.
• 🔌 Local tool servers (stdio). You can now connect MCP tool servers that run as local commands — not just over HTTP. Add the command and arguments from the Tool Servers tab.
• 📄 Read documents. The AI can now open and read PDFs, Word docs, Excel spreadsheets, PowerPoint files, and more.
• 💬 Send input to running commands. The AI can now type into running processes — answering prompts, interacting with REPLs, or sending Ctrl-C.
• ↩️ Undo last commit. Changed your mind? Undo the last commit from the Git history and get your changes back in staging.
• 🚀 Publish branches. Push a new branch for the first time with one click. The button says "Publish" when there's no upstream yet.
• 🔗 View on GitHub / GitLab. A new link in the Git panel takes you straight to your repo on the web.
• 📋 Commit actions menu. Click the dots on any commit to copy its hash. On the latest commit you can also undo it.
• 🔗 Share cptr. Quick links on the About page to share cptr on X, Reddit, LinkedIn, or copy the URL.

Changed

• ⚡ Smoother command execution. Commands now run in a real terminal (PTY) by default. You can choose how long to wait for output before moving on.
• 🧠 Better reasoning model support. Models like o3 and o4-mini now keep their chain of thought across tool calls, giving more accurate results.
• 📝 Updated README. New sections on accessing cptr from your phone and a list of compatible terminal agents.
• 🔧 Clearer error messages. Validation errors now show what actually went wrong instead of a generic status code.

Fixed

• 🔒 Gateway connections working again. Fixed an error that could break the gateway models endpoint.

v0.4.3

Changed

• 💬 Telegram rich message formatting. Telegram messages now use sendRichMessage / sendRichMessageDraft (Bot API InputRichMessage) for properly rendered Markdown with bold, italic, code blocks, and links. Falls back gracefully to plain text when the API is unavailable.
• 📝 Increased Telegram streaming buffer. The streaming buffer limit for Telegram was raised from 4 096 to 32 768 characters, allowing longer messages to stream without premature chunking.
• 🔒 Proxy middleware authentication. The reverse-proxy middleware now verifies the cptr_session cookie before forwarding requests, preventing unauthenticated access to proxied local services.
• 📖 README refresh. Updated the README with revised feature descriptions and setup instructions.

v0.4.2

Fixed

• 🔧 Tool calls no longer jump to the top during streaming. Fixed a regression from 0.4.1 where tool call indicators appeared at the very top of the assistant message while streaming, instead of in their correct position after the preceding text. They would snap back to the correct position only after the response completed. Caused by a premature text buffer flush in the usage event handler that was left over from the pre-parallel-execution refactor.

v0.4.1

Changed

• 🌍 i18n for tool servers, browser, and web settings. Replaced all remaining hardcoded English strings in the Tool Servers, Web/Browser, and admin panels with translation keys. Added ~60 new keys per locale across all 10 languages (de, en, es, fr, ja, ko, pt-BR, ru, zh-CN, zh-TW).
• 🎨 General settings layout refactor. Reorganised the General settings panel into a scrollable layout with grouped sections (Notifications, Updates, Message Queue). Webhook URL input is now full-width with a hint instead of an inline save button. The save button now shows a loading/saving state.
• 🖱️ Scrollbar-on-hover utility. Added a .scrollbar-hover CSS class that hides scrollbar thumbs until the user hovers, reducing visual clutter in scrollable panels.