Design for adding LlamaCppClient to support local LLM inference
via llama-server with CUDA. Target model: Qwen 2.5 14B Q4_K_M.
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Streaming responses (chunk-based)
- Rich markdown rendering with syntax highlighting
- Model switching via /model command
- Scrollable message history
- Enhanced status bar with token usage
- Inline error handling with auto-retry
- Feature parity between minimal and fullscreen modes
Covers:
- SessionManager for multi-frontend support
- Minimal readline TUI
- Fullscreen Ink-based TUI
- Mode switching and session transfer
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
8 tasks covering:
- New dependencies (sqlite, ollama, openai)
- OpenAI client for fallback
- Ollama client for local LLM
- Model router with fallback chain
- SQLite session persistence
- Hook engine for sensitive operations
- Telegram confirmation UI
- Full integration
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Separate sessions per frontend with explicit transfer
- Automatic backend routing based on task type
- Text responses by default (voice is future)
- Immediate notification delivery
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Initial design for a self-hosted personal AI agent with:
- Telegram + TUI frontends
- Multi-model routing with fallback chain
- Claude Code/OpenCode CLI integration
- Hook-based security for sensitive operations
- Tailscale-only network exposure
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
William Valentin