qualito 0.1.0

Quality metrics for AI-assisted development — measure and improve your Claude Code sessions

DQI — Delegation Quality Index

Measure and improve AI agent delegation quality. CLI tool + MCP server for Claude Code.

Installation

pip install dqi
# or with MCP server support:
pip install "dqi[mcp]"

MCP Server Setup

Add to your .claude.json or .mcp.json:

{
  "dqi": {
    "command": "uvx",
    "args": ["dqi-mcp"]
  }
}

Available Tools

Tool	Description
dqi_score	DQI score summary: average, trend, component breakdown, tier distribution
dqi_cost	Cost analysis: total spend, average per run, daily trend, waste estimate
dqi_patterns	Repeated task pattern detection with classification and recommendations
dqi_warnings	Underperforming workspace/task_type combos with actionable suggestions
dqi_templates	Task type inference and delegation template recommendations
dqi_incidents	Quality incidents: regressions, anomalies, severity tracking
dqi_slo	SLO compliance: quality, availability, and cost targets

Deployment

The DQI dashboard deploys as a split-stack: Railway (API) + Vercel (frontend).

API (Railway)

1. Connect the repo to Railway
2. Set environment variables:
   • DQI_JWT_SECRET — stable random secret for JWT signing
   • DQI_DIR — data directory path (e.g. /data/.dqi)
   • CORS_ORIGINS — comma-separated allowed origins (e.g. https://dqi.dev,https://www.dqi.dev)
3. Railway will auto-detect the Dockerfile and railway.json

Frontend (Vercel)

1. Set the root directory to src/dqi/dashboard/frontend
2. Set environment variable:
   • VITE_API_URL — Railway API URL (e.g. https://api.dqi.dev)
3. The vercel.json rewrites /api/* requests to the Railway backend

Local Development

# Copy and fill in env vars
cp env.example .env

# API (from repo root)
uv run uvicorn dqi.dashboard.app:create_app --factory --port 8090

# Frontend (from src/dqi/dashboard/frontend/)
npm install && npm run dev