Changelog.

Real-Time AI Streaming

• AI responses now appear directly in your note as they're being written — you can watch your content take shape in real time instead of waiting for the full result
• Works for all AI actions: insertions, rewrites, and selection replacements
• The AI modal stays minimal during generation — just "Writing..." with a cancel button — while content streams into the editor behind it

Read all the notes

Smarter AI Agent

• The AI chat agent is now significantly more reliable when editing your notes — tool calls are validated at the inference level, eliminating malformed or broken edits
• Added a new find-and-replace editing capability so the agent can make targeted changes to specific sections without affecting the rest of your document
• The agent now handles longer, multi-step tasks more gracefully with improved execution controls

Unified AI Experience

• The AI modal (Cmd+K) now works identically for both notes and notebooks — same interface, same streaming behavior, same keyboard shortcuts
• New "Open Chat Panel" button in the AI modal lets you seamlessly hand off a prompt to the full agent chat when you need a more conversational interaction
• Cleaner error messages when AI is busy — instead of a cryptic error, you'll see a friendly "AI is busy with another task" notice

Notebook Improvements

• Redesigned markdown cells with a structured header bar matching code cells — drag handles, cell indices, and inline toolbar controls are now consistent across cell types
• New notebook settings panel: configure run-all scroll speed and delay between cells
• Package manager now includes a "Delete All" button to quickly clear installed Python packages
• Smarter run-all execution: the view scrolls to each cell's output as it completes
• Consistent 1-based cell numbering across all cell types

Under the Hood

• Major internal cleanup: removed thousands of lines of legacy workaround code
• Improved tag cleanup when deleting notebooks — usage counts and orphan tags are now handled correctly
• Better Python kernel package handling with more reliable install defaults

Improvements

• More readable AI modals — The AI prompt panels (in both notes and notebooks) are now easier to read. The glassmorphic background is slightly more opaque so text and buttons no longer blend into the content behind them.

0.1.0-beta.6

Notebooks

• AI now sees your attached files. When you ask the notebook AI for help, it knows which files you've uploaded and can suggest the right code to use them (e.g. librosa.load('song.mp3'), pd.read_csv('data.csv')).
• AI knows how to save files back. The AI can now guide you through using save_to_notebook() to save outputs — DataFrames, JSON, images, or any file — back to the notebook as attachments.
• Fixed Run button animation. Running an individual cell now properly shows the spinning "Running..." indicator. Previously this only worked with Run All.
• Package manager now hints at version pinning. The install field placeholder shows numpy==1.26.4 as an example so you know you can pin specific versions.

Editor

• Fixed autocomplete overlapping with placeholder text. The ghost-text autocomplete no longer triggers when the cursor is on an empty line, preventing it from rendering on top of the "Start writing..." placeholder.

What's New

Fixed: Notes and notebooks failing to save on some machines

Some users upgrading from earlier versions of Canto were unable to create notes, daily notes, or notebooks. The app would show an error like "table has no column named..." — this was caused by a database upgrade step being silently skipped during the transition to encrypted storage. This has been fully resolved. Canto now automatically detects and repairs any missing database fields on startup, so affected users will be fixed immediately upon updating.

Improved: Notebook creation feedback

Creating a new notebook from the sidebar now shows a confirmation toast on success, or a detailed error message if something goes wrong — matching the behavior already in place for notes and daily notes.

What's New

You'll now see what's going on when something fails

Previously, if an operation failed (creating a note, saving a notebook, importing files, etc.), the app would silently do nothing — leaving you wondering what happened. Now, Canto shows clear notification messages whenever something goes wrong, including what the error actually was. This applies to:

• Creating, opening, and deleting notes
• Opening and saving notebooks
• Creating daily notes
• Creating folders
• Importing Markdown files
• Backing up and restoring your workspace
• AI model loading, downloading, and switching
• Attaching files to notes
• Searching notes
• Managing templates
• Moving your data directory
• Downloading app updates

You'll also see a confirmation message when operations succeed (e.g. "Note created", "Daily note opened", "Folder created"), so there's always clear feedback.

Improved app startup reliability

• The AI engine now starts up in the background after the app is fully loaded, rather than during startup. This prevents the AI initialization from potentially blocking note creation and other core features.
• Added a safety timeout so the AI engine can't get stuck loading forever.

Better error visibility for note creation

• If creating a note fails for any reason, the actual error message is now captured and displayed — making it much easier to diagnose and report issues.

What's New

🚀 Major Upgrade: New AI Engine

We've completely rebuilt Canto's AI engine from the ground up. The app now uses native Metal GPU acceleration on Apple Silicon Macs, delivering:

• Faster response times — AI responses start generating almost instantly
• Better memory efficiency — Smoother performance on all Mac configurations
• More reliable model switching — Seamlessly switch between AI models without restart
• Improved stability — No more "WebGPU not available" errors

🧠 Smarter Model Recommendations

The AI model picker now intelligently recommends the best model for your Mac:

• Automatically detects your RAM and CPU cores
• Shows a "Recommended" badge on the optimal model for your hardware
• Disables models that exceed your system's capabilities (prevents crashes)

✨ New AI Models

We've expanded our AI model lineup:

• Qwen 3 4B — Fast and efficient for everyday tasks
• Qwen 3 8B — The sweet spot for most users (recommended)
• Qwen 3 14B — Enhanced reasoning for complex work
• Qwen 3 32B — Professional-grade intelligence (for high-end Macs)
• GPT-OSS 20B — OpenAI's open research model
• Qwen3 Coder 30B — Specialized for coding tasks
• Qwen 3 Next 80B — Ultimate performance for power users (128GB+ RAM)

Removed: Qwen 3 1.7B (too limited), DeepSeek-R1 (no tool support)

🔄 Auto-Download on First Launch

When you open Canto for the first time (or switch models), it will automatically download your selected AI model in the background. No manual setup required!

🎨 UI Improvements

• Cleaner model cards — Better badges showing model capabilities (Thinking, Tools)
• Sleeker parameter tags — More readable model size indicators
• Consistent download buttons — Match the style of other action buttons
• Better download progress — Progress persists when you close and reopen Settings
• Fixed update notification padding — No longer overlaps with window buttons

🐛 Bug Fixes

• Fixed "No sequences left" error when using AI in multiple panes
• Fixed AI initialization errors when using presets or tag suggestions
• Fixed app quit handling to properly clean up GPU memory
• Fixed sidebar "Downloading model" text color (only the dot is orange now)

Read all the notes

System Requirements

• macOS 11 (Big Sur) or later
• Apple Siicon Mac (M1/M2/M3/M4) with at least 8GB RAM
• For best experience: 16GB+ RAM recommended

Native Python Execution

• Bundled Python runtime — Python code cells now run on a real CPython interpreter instead of the browser-based Pyodide. This means faster execution, full library compatibility, and no more WebAssembly limitations.
• Per-notebook isolation — Each notebook gets its own Python process, so variables and imports in one notebook never interfere with another.

Read all the notes

Package Management

• Install packages from code cells — Use !pip install pandas or %pip install numpy directly in a code cell, just like Jupyter. Packages are available immediately in the same cell.
• Package Manager — A new "Packages" button in the notebook toolbar opens a dedicated panel to browse, install, search, and uninstall Python packages with a visual interface.
• Global packages — Packages you install are available across all notebooks automatically.

Charts & Visualizations

• Inline matplotlib plots — Charts from plt.show() now render directly in the cell output instead of opening a separate window.
• HTML rendering — IPython.display.HTML(...) output now renders inline in the cell, enabling rich formatted output from Python cells.

Variable Sharing Across Languages

• Python ↔ JavaScript — Variables defined in a Python cell (strings, numbers, lists, dicts) are now accessible in JavaScript cells and vice versa. Mix languages freely within a single notebook.

File Attachments

• Attach files to notebooks — Drag and drop CSV, JSON, images, or any file into the new Files panel. Attached files are accessible from code cells automatically.
• Read files from code — Python cells can read attached files directly by filename. JavaScript cells access them via the __files__ object.
• Save files from code — Use save_to_notebook('output.csv', df) in Python or __saveFile__('data.json', text) in JavaScript to save results as notebook attachments.

Notebook AI

• AI modal — Press Cmd+K or click the sparkle button to open the notebook AI assistant with quick presets: Add Abstract, Add Conclusion, Suggest Next Step, Review Notebook, Visualize Data, Validate Data, and Consolidate Imports.
• Smarter notebook agent — The sidebar AI agent can now read full cell contents for more accurate answers about your code and data.

Notebook History

• Rich history preview — Version history now renders markdown cells with proper formatting and code cells with syntax highlighting, making it easier to compare versions.

Writing Style Profile

• Smarter regeneration — Your writing style profile no longer regenerates unnecessarily on every app launch. A new "Periodically Regenerate" toggle lets you control automatic updates.

UI & Polish

• Compact notebook toolbar — Toolbar buttons are now icon-only with tooltips for a cleaner, more spacious layout.
• Updated pricing — Canto is now $14.99 (down from $24.99).

New Features

Semantic Connections in Knowledge Graph

The knowledge graph can now visualize semantic relationships between your notes. Canto analyzes your note content using AI embeddings and draws connections between notes that are conceptually similar — even if they don't directly link to each other. Toggle semantic edges on or off with the new pink "Semantic" filter in the graph toolbar.

Inline Folder Renaming

You can now rename folders directly in the file tree — just right-click a folder and choose "Rename." No more modal dialogs; the name becomes editable right where it is.

Backup Size Tracking

The Backups settings page now shows the total size of all your backup snapshots (in KB, MB, or GB), so you can keep an eye on disk usage at a glance.

New Keyboard Shortcuts

• ⌘ + Option + N — Create a note from a template
• ⌘ + Shift + T — Toggle dark/light mode
• ⌘ + Shift + D — Delete the active code cell in a notebook

---

Read all the notes

Improvements

Full Light Theme Support

The light theme now works correctly across the entire app:

• Knowledge Graph — All backgrounds, text, borders, and node labels properly adapt to light mode
• Notebook Code Cells — The code editor switches between dark and light themes automatically
• Notebook Markdown Cells — Rendered markdown text is now visible on light backgrounds (previously white-on-white)

Redesigned Backup Settings

The Backups settings page has been completely reorganized into a cleaner, more compact layout:

• All settings visible without scrolling
• Status info (last backup, next scheduled, total size, health) shown in a single line
• Toggle switches replace checkboxes for encryption and embeddings options
• "View backups & restore" is now a prominent action button

Improved Knowledge Graph Toolbar

The graph toolbar has been restructured into a two-row layout with the search bar on its own row. Filter toggles, layout buttons, and zoom controls are better organized and easier to use.

Better Tab and Layout Behavior

• Tabs now distribute width more naturally based on available space
• Editor and toolbar padding is more consistent across the app
• The secondary sidebar integrates more smoothly with the main layout
• Traffic light buttons (close/minimize/maximize) are better aligned with the tab bar

Cleaner Pane Creation

Opening a new split pane no longer creates a default empty tab — panes start clean, ready for whatever you want to open in them.

---

Bug Fixes

• Fixed markdown cells showing invisible (white on white) text in light mode
• Fixed knowledge graph staying dark when the app theme was set to light
• Fixed code cells always using the dark editor theme regardless of app settings
• Fixed inconsistent padding on sidebar buttons and editor toolbar
• Fixed tab width calculation that could cause layout overflow