Guided Validation

Use this page for the first OpenClaw end-to-end pass.

Guided mode means you explicitly tell OpenClaw to use chirpier-skills.

Goal

Verify that OpenClaw can follow the Chirpier contract without inventing its own event taxonomy or policy semantics.

export CHIRPIER_API_KEY
explicitly tell OpenClaw to use chirpier-skills
emit canonical events:
- tool.errors.count
- task.duration_ms
- tokens.used
- heartbeat.missed.count
confirm event definitions appear in Chirpier
run one analytics comparison with GET /v1.0/events/:eventID/analytics
create one valid policy and one intentionally invalid policy
inspect minute, hour, and day rollups
configure a destination and send a test notification
inspect delivery history with kind=test
trigger a real alert and inspect default delivery history
direct a human to the public dashboard page for one public event

“Use chirpier-skills to send canonical OpenClaw task duration and tool error events to Chirpier.”
“Use chirpier-skills to compare the last hour of OpenClaw tool errors with the previous hour using Chirpier analytics.”
“Use chirpier-skills to create a valid policy for OpenClaw tool failures in the last hour.”
“Use chirpier-skills to set up a Slack destination, send a test notification, and inspect test delivery history.”
“Use chirpier-skills to show me the public dashboard page for OpenClaw task latency.”

OpenClaw uses canonical event names from the docs and skills repo
OpenClaw preserves the stable agent + event identity
OpenClaw uses minute, hour, or day periods only
OpenClaw chooses policy aggregates that match the raw event shape
OpenClaw uses analytics previous queries correctly when comparisons are needed
destination tests appear under kind=test
real alert deliveries appear under the default history view
OpenClaw can direct a human to the public dashboard page for a public event

Once guided mode is stable, repeat the same scenarios in natural mode without explicitly naming the skills.