One-command startup for each repo's frontend + backend + browser.
Each repo has a startup script that automatically launches the backend, frontend, and opens your browser:
# Start any individual repo
cd squish && ./scripts/start.sh
cd kyro && ./scripts/start.sh
cd kairu && ./scripts/start.sh
cd vectro && ./scripts/start.sh
cd kohaku && ./scripts/start.sh
cd toki && ./scripts/start.shOr use the master script from the repo root:
# Start individual repos
./scripts/start-konjo.sh vectro
./scripts/start-konjo.sh kairu kohaku
# Start all repos
./scripts/start-konjo.sh all- Kills existing processes on the target ports (prevents port conflicts)
- Starts the backend (FastAPI or demo server)
- Waits for backend to be ready (polls
/healthendpoint) - Starts the frontend dev server (Vite on 517X)
- Waits for frontend to be ready (polls
http://localhost:517X) - Opens your browser automatically
- Shows summary with all running ports
- Manages cleanup on Ctrl+C
| Repo | Backend | Frontend | Notes |
|---|---|---|---|
| squish | 11435 (FastAPI) + 8001 (demo) | 5177 | Both servers for comparison feature |
| kyro | 8000 (FastAPI) + 8766 (demo) | 5178 | RAG pipeline visualization |
| kairu | 7777 (demo) | 5176 | Speculative decoding |
| vectro | 8765 (demo) | 5179 | Embedding compression |
| kohaku | 8000 (demo) | 5180 | Hypervector memory |
| toki | 8765 (demo) | 5181 | Adversarial hardening |
Note: kohaku and kyro both use port 8000 (can't run simultaneously), and vectro/toki both use 8765 (can't run simultaneously).
Each script logs backend and frontend output to /tmp/:
# View logs while running
tail -f /tmp/vectro_backend.log
tail -f /tmp/vectro_demo.log
tail -f /tmp/vectro_frontend.logAnother instance is running. Kill it:
pkill -f "python.*demo"
pkill -f "vite"The Vite dev server takes longer to start on first run. Wait another 30 seconds, or check logs:
tail -f /tmp/{repo}_frontend.logCheck if dependencies are installed or backend code has issues:
# For FastAPI repos (squish, kyro)
python -m pip install -e .
python -m konjoai.api.app --port 8000
# For demo servers
cd {repo}/demo && python server.py --port 8765Make sure scripts are executable:
chmod +x squish/scripts/start.sh
chmod +x ./scripts/start-konjo.shIf the scripts fail, start manually in separate terminals:
# Terminal 1: Backend
cd vectro
python demo/server.py --port 8765
# Terminal 2: Frontend
cd vectro/dashboard
npm run dev
# Terminal 3: Browser
open http://localhost:5179Edit the port numbers in each scripts/start.sh if you need different ports:
FRONTEND_PORT=5179
DEMO_PORT=8765Or temporarily override with environment variables (modify script to support this).
Quick reminder: These scripts assume Python 3.10+, Node 18+, and npm/Node dependencies are installed.