Keyboard shortcuts and commands
Every BdebTech AI command, its default keybinding, and the settings that control autocomplete and chat.
Shortcuts
| Action | Windows / Linux | macOS |
|---|---|---|
| Accept autocomplete suggestion | Tab | Tab |
| Accept next word of suggestion | Ctrl+→ | ⌘+→ |
| Dismiss suggestion | Esc | Esc |
| Open chat | Ctrl+Alt+L | ⌘+⌥+L |
| Edit selection with AI | Ctrl+Alt+K | ⌘+⌥+K |
| Plan and edit (multi-file agent) | Ctrl+Alt+Shift+K | ⌘+⌥+⇧+K |
| Toggle tab autocomplete | Ctrl+Alt+Shift+A | ⌘+⌥+⇧+A |
Change any of them under File → Preferences → Keyboard Shortcuts, search “BdebTech”.
Commands (Command Palette, Ctrl+Shift+P)
- BdebTech AI: Open Chat / New Chat
- BdebTech AI: Edit Selection with AI
- BdebTech AI: Plan and Edit (multi-file agent) - describe a change; the agent picks up to 6 files, proposes complete rewrites, and you apply all, review each diff, or discard. Needs Lite+ or a Studio model; output is schema-constrained JSON so it never half-writes a file.
- BdebTech AI: Index Workspace for Repository-Aware AI / Clear Workspace Index
- BdebTech AI: Start Free Studio Trial (7 days) / Buy Studio License
- BdebTech AI: Explain Selection / Fix / Improve Selection / Generate Tests for Selection
- BdebTech AI: Switch Model
- BdebTech AI: Download Starter Model / Run Setup
- BdebTech AI: Toggle Tab Autocomplete
- BdebTech AI: Restart Local AI Engine / Restart Local AI Daemon
- BdebTech AI: Show Status Menu (also by clicking the status bar item)
- BdebTech AI: Show Logs
- BdebTech AI: Activate Studio License
Settings
| Setting | Default | What it does |
|---|---|---|
bdebtech.completions.enabled | true | Tab autocomplete on/off |
bdebtech.completions.debounceMs | 220 | Wait after typing stops before asking the model |
bdebtech.completions.maxTokens | 64 | Length cap per suggestion |
bdebtech.completions.multiline | auto | always, never, or multi-line only at line end |
bdebtech.completions.contextChars | 3000 | Code before the cursor sent as prefix (suffix gets half) |
bdebtech.completions.disabledLanguages | plaintext, markdown, log | Where autocomplete stays quiet |
bdebtech.chat.temperature | 0.2 | Lower = more deterministic |
bdebtech.chat.maxTokens | 1024 | Reply length cap |
bdebtech.chat.maxContextChars | 12000 | File/selection context budget |
bdebtech.chat.systemPrompt | empty | Extra standing instructions |
bdebtech.daemon.url | http://127.0.0.1:41337 | Where the daemon listens |
bdebtech.daemon.path | auto | Explicit bdebd location |
bdebtech.daemon.autoStart | true | Start the daemon with the editor |
Tips
- Autocomplete is single-line when there is code after the cursor on the same line, and multi-line at line ends. Press Enter then pause to get a whole block.
- Write the function signature and a one-line comment first; small models complete intent far better than they guess it.
- In chat, select the exact lines you are asking about instead of attaching the whole file; the answer is faster and more precise.