Documentation Index
Fetch the complete documentation index at: https://docs.selftune.dev/llms.txt
Use this file to discover all available pages before exploring further.
Overview
Alert rules notify you when skills experience regressions, performance drops, or data gaps. Configure multiple alert types and delivery channels to stay informed about skill health in real time.Alert Types
selftune supports four alert types, each with distinct thresholds:| Type | Trigger | Threshold | Use Case |
|---|---|---|---|
| Pass rate drops below | Skill pass rate falls below a percentage | 0–100% | Catch quality regressions |
| Regression detected | New evolution performs worse than baseline | (no threshold) | Monitor post-deploy safety |
| No data received | Skill hasn’t run in N days | Days (1–90) | Detect stalled deployments |
| Evolution failed | Autonomous evolution attempt failed | (no threshold) | Track evolution issues |
Notification Channels
Configure alerts to deliver via your preferred channel:Slack
Post alerts to a Slack channel via webhook. Requires a Slack workspace and webhook URL from your Slack app settings.Discord
Post alerts to a Discord channel. Similar to Slack, requires a webhook URL.Webhook
Send POST requests to a custom endpoint. selftune POST’s the alert payload to your URL.Creating Alerts
Dashboard UI
- Navigate to Settings → Alerts
- Click + Create Rule
- Select alert type and configure threshold (if applicable)
- Choose notification channel and enter configuration
- Click Save Rule
API
Create alerts programmatically:Managing Alerts
- Enable/disable — toggle the rule on without deleting
- Delete — remove the rule permanently
- Edit — currently, delete and recreate with new settings
Best Practices
- Critical alerts only — set high thresholds to avoid noise (e.g., pass_rate_below 60%, not 95%)
- Multiple channels — create separate rules for different channels and severities
- No data thresholds — set 7–14 days to allow for natural variation
- Test webhooks — validate custom webhook URLs before deploying to production