Starter Dashboards

The fastest way to make OpenClaw useful in Chirpier is to build one small dashboard that covers reliability, latency, cost, and liveness, then hand people the public dashboard page for the trends they care about.

Recommended first dashboard

Start with these charts:

tool.errors.count by hour with sum
task.duration_ms by hour with average
task.duration_ms by day with p95_est
tokens.used by hour with sum
heartbeat.missed.count by day with sum

For v1, treat meta as inspection-only context. Prefer stable event naming and chart configuration choices rather than relying on meta as a first-class analytical filter.

Why this works

This layout gives you a strong first pass across:

reliability: tool failures
latency: task duration
cost: token usage
liveness: missed heartbeats

Build this before creating lots of policies. Good dashboards make alert tuning easier.

Analytics vs dashboard

use analytics when OpenClaw needs a direct period-vs-previous comparison
use dashboards when a human needs to inspect multiple trends together
use the public dashboard page when you want to hand one event trend directly to a human

Useful follow-up dashboards

per-tool reliability dashboard
per-workflow latency dashboard
sentiment or market monitoring dashboard
destination health and alert activity dashboard

Period contract

For backend rollups and alerting, use the canonical periods:

minute
hour
day

Frontend date-range shortcuts like 30d are chart-layer concepts, not replacements for the backend rollup period enum.

Public dashboard page

When an event is public, direct users to the public dashboard page so they can inspect the trend directly. Use Public Event Pages for the hosted page contract.