Key Features
Multiple Provider Support
Connect to leading AI providers:- OpenAI - GPT-4, GPT-4o, GPT-4 Turbo, and more
- Anthropic - Claude 3.5 Sonnet, Claude 3 Opus, Haiku
- Google Gemini - Gemini Pro, Gemini Flash
- OpenRouter - Access 100+ models through a single API
- Ollama - Run models locally on your machine
- OpenAI-compatible - Any OpenAI-compatible endpoint
- Codex (ChatGPT OAuth) - ChatGPT integration via OAuth authentication
Profile Management
Create multiple AI profiles with different configurations:- Switch between providers and models instantly
- Configure model-specific parameters like reasoning effort
- Store API keys securely per-space in
.glyph/app/ai_secrets.json - Set default profiles for quick access
Context Attachment
Ground AI responses in your notes:- Attach individual files or entire folders to conversations
- Mention files with
@filenamesyntax in your prompts - Configure character budgets (200-250,000 chars)
- View token estimates before sending
- Context is stored per-conversation for repeatability
Chat vs Create Modes
Chat Mode - Conversational interaction without tools- Back-and-forth dialogue with the model
- No file system access
- Faster responses
- Best for questions, brainstorming, and discussion
- Access to file reading, searching, and listing tools
- AI can explore your notes to answer questions
- Tool usage tracked in timeline view
- Best for research, summarization, and knowledge work
History and Audit Trail
Every conversation is automatically saved:- History stored in
.glyph/ai_history/as JSON - Review past conversations and their context
- Tool usage events logged per conversation
- Export conversation data for compliance or analysis
Quick Start
Choose a Provider
Go to Settings → AI and select a provider from the available profiles. Each profile supports a different AI service.
Configure Authentication
Add your API key or authenticate via OAuth (for Codex). Keys are stored securely per space.
Select a Model
Choose from available models for your provider. Glyph fetches the latest model list from each API.
Architecture
- Frontend: React components in
src/components/ai/ - Backend: Rust
ai_rigmodule insrc-tauri/src/ai_rig/ - AI Framework: Rig for unified provider interface
- Storage: SQLite index + JSON files in
.glyph/ - Tools: Space-scoped filesystem operations (read, search, list)
Security
- API keys stored encrypted in space-local
ai_secrets.json - Tool operations restricted to current space (no path traversal)
- SSRF protection for custom base URLs
- Optional
allow_private_hostsflag for local models - No hidden file access (dotfiles blocked)
- Request/response audit trail available
Next Steps
Setup
Configure your first AI profile
Providers
Provider-specific setup guides
Chat Modes
Learn about chat vs create modes
Context Management
Attach notes and folders to conversations