A real-time operations dashboard for OpenClaw — monitor your AI agents, track costs, manage cron jobs, and watch system health from one interface.
The easiest way is to tell your OpenClaw agent:
"Install the openclaw-dashboard skill"
OpenClaw will automatically install and configure the dashboard. After installation, start the server:
cd ~/.openclaw/workspace/skills/openclaw-dashboard
cp env.example .env
# Edit .env — set OPENCLAW_AUTH_TOKEN
node api-server.jsOpen http://localhost:18791/ in your browser.
git clone https://github.com/JonathanJing/openclaw-dashboard.git
cd openclaw-dashboard
cp env.example .env
# Edit .env — set OPENCLAW_AUTH_TOKEN at minimum
node api-server.jsOpenClaw Dashboard gives you a live window into your running OpenClaw gateway. It reads session logs, cron history, and watchdog state directly from ~/.openclaw/ and presents them in five panels.
What you see:
- 5 stat cards: today's cost, total tokens, active cron jobs, active sessions, model mix breakdown
- Today's Cron Runs — collapsible log of every scheduled job with model, tokens, cost, and time
- System info bar — hostname, macOS, CPU, load, RAM %, disk %, Node and OpenClaw versions
- Session alerts — flags over-spec (wasting money) or under-spec sessions automatically
- Sessions table — every active channel with model selector, message count, tokens, cost, $/msg, and fit score
- Language toggle (top-right
ENbutton) — switch entire UI between English and Chinese
What you see:
- Today's usage with model breakdown pills
- Channel Breakdown — per-channel token and dollar spend
- All-Time Usage — lifetime totals by model (Opus / Sonnet / Flash / GPT / Gemini)
- Daily Tokens chart — 14-day stacked bar chart by model
- Daily Cost chart — dollar spend per day by model
- Cost Heatmap — model × day matrix showing where spend concentrates
What you see:
- All configured cron jobs with schedule, description, last run status, and model
- Inline model selector — change the LLM for any job without editing config files
- Cost Analysis — per-job lifetime cost, avg tokens/run, avg $/run, today's spend
- Run History — per-job daily breakdown for last 7 days
- Trend chart — fixed baseline vs. variable cost over time
What you see:
- Session Quality — silence rate per channel (NO_REPLY + HEARTBEAT_OK / total). High silence = consider downgrading model to save cost
- Audit — live provider API status (OpenAI / Anthropic / Google) with request counts
- Config Audit — per-channel model recommendations with over/under-spec flags
- System Info — full system details + dashboard uptime
- Operations — one-click Backup + Push, Restart (when enabled via env)
- Watchdog — live gateway health with:
- Status pills: Running state / failure count / config drift detection
- 24h uptime bar — color-coded blocks (green=healthy, yellow=degraded, red=down)
- Incident log — grouped outage events with duration, reason, and recovery time
Copy env.example to .env and edit.
| Variable | Default | Description |
|---|---|---|
OPENCLAW_AUTH_TOKEN |
(none) | Dashboard access token. Always set this. |
DASHBOARD_HOST |
127.0.0.1 |
Bind address. Keep localhost unless using a tunnel. |
DASHBOARD_PORT |
18791 |
Port to serve on. |
| Variable | What it enables |
|---|---|
OPENCLAW_ENABLE_MUTATING_OPS=1 |
Backup, restore, update, and model-change actions |
OPENCLAW_LOAD_KEYS_ENV=1 |
Load ~/.openclaw/keys.env on startup |
OPENCLAW_ENABLE_PROVIDER_AUDIT=1 |
Fetch live usage from OpenAI / Anthropic org APIs |
OPENCLAW_ENABLE_CONFIG_ENDPOINT=1 |
Expose /ops/config read endpoint |
OPENCLAW_ENABLE_SYSTEMCTL_RESTART=1 |
Allow user-scoped systemctl restart |
If exposing beyond localhost (e.g. Tailscale Funnel), always set auth token and CORS:
OPENCLAW_AUTH_TOKEN=your_token_here
DASHBOARD_CORS_ORIGINS=https://your-tailscale-host.ts.net- Binds to
localhostby default - Auth stored in HttpOnly cookie after login — not in URL params
- All subprocess calls use
execFileSyncwith args array (no shell injection) - Mutating operations require
OPENCLAW_ENABLE_MUTATING_OPS=1AND a localhost origin - Never commit
.envor your auth token
| File | Description |
|---|---|
agent-dashboard.html |
Single-file frontend |
api-server.js |
Backend API server |
SKILL.md |
OpenClaw agent install instructions |
env.example |
Environment variable template |
MIT — built by Jony Jing



