Self-Healing
APIs change. Fields get renamed, schemas evolve, new required fields appear, and deprecated endpoints get removed. Traditional integration platforms break silently when this happens. fyrn detects these changes and fixes your mappings automatically.
How it works
Section titled “How it works”The self-healing engine runs continuously and follows a four-step process:
1. Monitor
Section titled “1. Monitor”fyrn monitors every connected API for schema changes. This includes:
- Field renames (e.g.,
unit_amount→amount_per_unit) - New required fields
- Type changes (e.g., string → integer)
- Deprecated fields and endpoints
- API version updates
Monitoring happens through a combination of runtime response analysis and periodic schema polling.
2. Detect
Section titled “2. Detect”When a schema change affects one of your flows, the engine creates a drift event. The event captures:
- Which connector and API version changed
- Which fields were affected
- Which flows reference those fields
- The exact nature of the change (rename, addition, removal, type change)
Detection typically occurs within 2 seconds of the first divergent API response.
3. Analyze
Section titled “3. Analyze”The AI analyzes the drift event and generates a proposed fix:
- For field renames: maps the old field to the new one
- For new required fields: suggests a default value based on the field type and context
- For type changes: adds type coercion to the transform step
- For removed fields: flags for human review (no safe auto-fix)
Each proposed fix receives a confidence score from 0-100%.
4. Fix or review
Section titled “4. Fix or review”The confidence threshold (configurable per flow, default: 95%) determines what happens next:
| Confidence | Action |
|---|---|
| ≥ 95% | Auto-applied. Flow updated, logged, notification sent. |
| 80-94% | Routed for human review. Suggested fix provided. |
| < 80% | Flagged as manual. Requires developer intervention. |
Configuration
Section titled “Configuration”Configure self-healing per flow:
healing: auto_fix: true # Enable/disable auto-fixing confidence_threshold: 0.95 # Minimum confidence for auto-fix notify_on_fix: true # Send notification on every fix review_channel: slack # Where to send review requestsOr set a global default in your project config:
healing: default_threshold: 0.95 global_notify: trueAudit trail
Section titled “Audit trail”Every self-healing event is logged:
# View recent healing eventsfyrn healing log
# View events for a specific flowfyrn healing log --flow shopify-to-sap-ordersEach log entry includes the drift event, the proposed fix, the confidence score, and whether it was auto-applied or sent for review.
Next steps
Section titled “Next steps”- Governance — Policy controls for self-healing behavior
- Architecture — How self-healing fits into the platform