mirror of
https://github.com/getpaseo/paseo.git
synced 2026-10-23 17:52:49 +00:00
f3b0199c13
* fix(server): skip Claude subagent replay over a transcript size budget Loading a session's subagent transcripts all at once OOMs the daemon worker on large Claude sessions (#5820: a 395 MB parent plus 269 sidechain transcripts, 757 MB total, crash-loops the worker). Stat the sidechain files before reading any and leave them unread when the parent plus sidechain transcripts exceed a replay budget (256 MB by default, PASEO_CLAUDE_REPLAY_BUDGET_MB to override); the parent history replays unchanged, which is the same state as moving the subagent transcripts out of the way. Also parse the parent transcript once instead of twice while rebuilding sidechain ownership. * fix(server): accept fractional PASEO_CLAUDE_REPLAY_BUDGET_MB values parseInt read 0.5 as 0, so a fractional budget silently fell back to the 256 MB default — the opposite of what a small-host operator asked for. Parse the value as a float and pin the behavior with a regression test. * Stream Claude transcript replay without discarding subagent history Read JSONL incrementally, retain ownership facts instead of raw records, and replay workflow payloads through a chronological byte-location index. Preserve all subagent history and remove the input byte budget. Co-authored-by: ZouYusheng <admin@tj7mun.org> * fix(server): propagate Claude transcript replay read errors * fix(server): bound Claude replay output and index legacy history Apply the existing tool content limit before replay retains events, copying truncated prefixes so sliced strings do not pin the original output. Index each child once and own transcript snapshots through ingestion to preserve history across provider file deletion. Keep full-scale memory reproductions opt-in. * fix(server): replay Claude history without temporary transcript copies Keep the original descriptors open until synchronous replay finishes. Indexed passes survive unlink and rename without requiring a second copy of the session on temporary storage. Close every descriptor on success and failure. * fix(claude): replay large child histories without spread Streaming removes whole-transcript duplication, but folding many visible child events still passed the complete aggregate as arguments to Array.push. Large histories exceed the V8 argument limit, the loader catches a RangeError, and history remains empty. Append replay observations iteratively and cover 160,000 aggregate child events. Adapted from #7511, independently reproduced with the public resumeSession and streamHistory interfaces. Co-authored-by: Alin Jiang <zonglinj@amazon.com> * test(claude): cover parent-only replay above the string limit Exercise the same parent-only resume path routinely with a small corpus, and use PASEO_TEST_LARGE_CLAUDE_REPLAY=1 for the 550 MiB string-limit reproduction. The eager loader returns empty history for this corpus; incremental replay preserves the visible parent event. --------- Co-authored-by: Yusheng Zou <admin@tj7mun.org> Co-authored-by: Mohamed Boudra <boudra.moha@gmail.com> Co-authored-by: paseo-bot[bot] <266920839+paseo-bot[bot]@users.noreply.github.com> Co-authored-by: Alin Jiang <zonglinj@amazon.com>
Voice Assistant
A voice-controlled terminal assistant that runs as a single local service.
Quick Start
# Install dependencies
npm install
# Copy environment variables
cp .env.example .env
# Edit .env and add your API keys (OpenAI, Deepgram)
# Run development servers
npm run dev
# Open browser to http://localhost:5173
Architecture
- Express Server (port 3000) - Serves API and built UI in production
- Vite Dev Server (port 5173) - Hot-reload React UI in development
- WebSocket (
/ws) - Real-time bidirectional communication - Agent - STT → LLM → TTS pipeline with terminal control
- Daemon - tmux-based terminal management (in-process)
Development
# Run both servers (recommended)
npm run dev
# Or run separately:
npm run dev:server # Express on port 3000
npm run dev:ui # Vite on port 5173
# Type checking
npm run typecheck
# Build for production
npm run build
# Start production server
npm start
Project Status
✅ Completed (Phases 1-2):
- Package setup and configuration
- Express server with WebSocket
- React UI with Vite
- WebSocket client with ping/pong testing
⏳ In Progress (Phase 3):
- Terminal control (tmux integration)
📋 Planned (Phases 4-9):
- LLM integration (OpenAI GPT-4)
- Agent orchestrator
- Speech-to-Text (Deepgram)
- Text-to-Speech (OpenAI)
- Audio streaming
- UI polish
See IMPLEMENTATION_PLAN.md for complete details.
Environment Variables
OPENAI_API_KEY=your-openai-key-here # GPT-4 and TTS
DEEPGRAM_API_KEY=your-deepgram-key-here # Streaming STT
STT_MODEL=whisper-1 # Optional: override to gpt-4o-transcribe, etc.
STT_CONFIDENCE_THRESHOLD=-3.0 # Optional: reject low-confidence clips
STT_DEBUG_AUDIO_DIR=.stt-debug # Optional: persist raw dictation audio for debugging
PASEO_HOME=~/.paseo # Runtime state directory (agents/, etc.)
PASEO_LISTEN=127.0.0.1:6767 # Listen address (host:port or /path/to/socket)
PASEO_HOME defaults to ~/.paseo and isolates runtime artifacts like agents/. PASEO_LISTEN controls the daemon listen address. For blue/green testing you can run a parallel server without touching production state:
PASEO_HOME=~/.paseo-blue PASEO_LISTEN=127.0.0.1:7777 npm run dev
Tech Stack
- Server: Express, TypeScript, ws (WebSocket)
- Client: React 18, Vite, TypeScript
- Terminal: tmux (via child_process)
- AI: OpenAI (LLM + TTS), Deepgram (STT)
Testing
Currently manual testing via:
- Start servers:
npm run dev - Open http://localhost:5173
- Test WebSocket connection (green status indicator)
- Click "Send Ping" button to test communication
More testing guidance as features are implemented.
License
MIT