[🐛 Bug]: Incidents do not enrich inside workflow step

[PaulieNor]

Describe the bug

When running this workflow:

workflow:
  id: id
  name: Alert
  description: Alert
  disabled: false
  triggers:
  - type: incident
    events:
    - created
  inputs: []
  owners: []
  services: []
  consts: {}
  steps: []
  actions:
  - name: create-jira-ticket-keep-notification-board
    foreach: "{{ incident.services }}"
    if: '{{ foreach.value }} == "service"'
    provider:
      type: jira
      config: "{{ providers.Jira }}"
      with:
        board_name: Board Name
        issue_type: Task
        summary: Summary
        project_key: KE
        labels:
        - keep
        - data-notifications-{{ foreach.value }}
        description: '
          '
        enrich_incident:
        - key: ticket_url
          value: results.ticket_url

I get this error:

2026-06-24T11:53:04.808000: Running create-jira-ticket-keep-notification-board StepType.ACTION, current retry: 0
2026-06-24T11:53:04.809000: Notifying jira...
2026-06-24T11:53:04.809000: Creating an issue...
2026-06-24T11:53:05.957000: Created an issue!
2026-06-24T11:53:05.959000: Notified jira!
2026-06-24T11:53:05.959000: No fingerprint found for alert enrichment
2026-06-24T11:53:05.959000: No fingerprint found for alert enrichment
2026-06-24T11:53:05.959000: Failed to run step create-jira-ticket-keep-notification-board with error No fingerprint found for alert enrichment

However the fingerprint does exist in the workflow as it is triggered by an incident.

To Reproduce

Run a workflow created by an incident which uses a foreach loop, and inside the for_each loop try to enrich the incident.

Expected behavior

Incident should be enriched with correct information instead of erroring.

Additional context
According to Copilot, the issue is in _enrich() in base_provider.py. There is a step which is unreachable under the above conditions:

elif self.context_manager.foreach_context.get("value", {}):
    # enters here because foreach value is "customer" (truthy)
    foreach_context = ...
    if isinstance(foreach_context, AlertDto): ...       # No - it's a string
    elif isinstance(foreach_context, dict) and ...: ... # No - it's a string  
    elif self.context_manager.event_context: ...        # No - incident trigger
    else:
        fingerprint = None  # ← stuck here

# This incident_context branch is NEVER reached:
elif self.context_manager.incident_context:
    entity_type = "incident"
    fingerprint = self.context_manager.incident_context.id

[dosubot]

Your analysis is correct. This is a real bug in the _enrich method in base_provider.py [1].

The problem is at line 212: elif self.context_manager.foreach_context.get("value", {}) — when iterating over {{ incident.services }}, each foreach value is a string (e.g., "service"). Since a non-empty string is truthy, this branch is entered. Then:

1. isinstance(foreach_context, tuple) → ❌ (it's a string)
2. isinstance(foreach_context, AlertDto) → ❌ (it's a string)
3. isinstance(foreach_context, dict) and "fingerprint" in foreach_context → ❌ (it's a string)
4. self.context_manager.event_context → ❌ (incident trigger, no event context)
5. Falls to else → fingerprint = None → raises the exception

The incident_context branch at line 245 is never reached because the foreach string value already matched the condition at line 212. The fix would need to either:

• Add a check for incident_context inside the foreach branch (before the else at line 229), e.g.:

  elif self.context_manager.incident_context:
      entity_type = "incident"
      fingerprint = self.context_manager.incident_context.id

• Or restructure the priority so that when foreach value isn't an alert-like object and there's no event_context, it falls through to check incident_context.

This is similar in nature to a previously fixed bug in PR #6213 where foreach handling didn't properly set the event variable for AlertDto paths. The pattern of foreach branches not accounting for all value types in combination with incident triggers is a recurring gap in this logic.

To reply, just mention @dosu.

---

Share context across your team and agents. Try Dosu.

[archeems]

Hey there! I'd love to take a look at this for my first contribution. Could you please assign it to me?