Errors - Failproof AI

Errors gives you a failure-first view across sessions. Group by error type, agent, environment, model, tool, or time window to find repeated operational problems.

Investigate errors

Dashboard
CLI

Go to Observe → Errors.
Filter by environment, event type, error type, agent, session ID, or search text.
Expand a grouped error to see occurrences. Select a row to open the exact event in its session.
Select the bell control on a representative error to start an alert for similar errors.

fp errors --env production --since 24h
fp errors --error-type TimeoutError --agent-id checkout-agent
fp errors --aggregate --env production --since 7d

Use fp events --full --session-id <id> when you need the raw payload behind an error summary.

Use Errors when you need to

Find the most frequent error class in production.
Determine whether one tool or model causes a spike.
Jump from an aggregate count to representative sessions.
Create an alert for recurrence.
Include the affected population in an audit.

An error is an observed event. A failed evaluation is a quality judgment, and an audit finding is an investigated failure pattern. Keep those distinctions when deciding which response workflow to use.

Create an alert

Notify responders when an error or quality condition crosses a threshold.

​Investigate errors

​Use Errors when you need to

Create an alert

Investigate errors

Use Errors when you need to