Error handling in workflows is undocumented. We need a dedicated docs section covering:
- Output channels
- How node failures propagate through a canvas
- Configuring retry policies
- Using conditional edges to handle errors (e.g. on-failure branches)
- Stopping vs continuing a run on error
- Surfacing error details in downstream nodes via expressions
- Common patterns (e.g. retry with backoff, fallback path, error notification)
Error handling in workflows is undocumented. We need a dedicated docs section covering: