OpenClaw Usage Dashboard

Generate a self-contained HTML dashboard from OpenClaw session transcripts. Zero dependencies — Python 3.8+ stdlib only.

Usage

# Real data (reads ~/.openclaw/agents/main/sessions/)
python3 scripts/usage-dashboard-generic.py [output.html]

# Demo mode (generates 30 days of realistic mock data)
python3 scripts/usage-dashboard-generic.py [output.html] --demo

Default output: ~/openclaw-usage-dashboard.html

Override sessions path: set OPENCLAW_ROOT env var.

What It Shows

• Cost KPIs: Total spend, today's cost, API calls, active model, cache hit rate, monthly projection
• Model config: Active / default / fallback models from openclaw.json
• Anthropic quota gauges: 5h and 7d rate limit utilization (requires ANTHROPIC_API_KEY)
• Donut chart: Cost distribution by model
• Hourly sparkline: Today's call volume by hour
• Stacked bars: Cost by period (daily / weekly / monthly / yearly)
• Model table: Per-model breakdown with token counts and cost
• Detail table + chart: Toggle between table and bar chart view per period
• Budget tracker: Set monthly limit, progress bar, localStorage persistence
• CSV export: Download any period view as CSV
• Provider dropdown: Filter by provider

Supported Providers

Anthropic, OpenAI, Google, Moonshot/Kimi, Mistral, Cohere, Groq — 30+ models with built-in pricing.

Customization

Add models

Edit PRICES dict in the script (price per 1M tokens):

"provider/model-name": {"input": 2.50, "output": 10.00, "cache_read": 1.25, "cache_write": 2.50}

Add to MODEL_COLORS and ALIAS_MAP as needed.

Pricing reference

See references/prices.md for current prices across all providers.

Requirements

• Python 3.8+ (stdlib only)
• ANTHROPIC_API_KEY in env (optional, for live quota gauges)