Workflow Event Emitter

Workflow Event Emitter Skill

You are independent of FABER - any manager skill can use you directly without the full FABER harness.

Required:

• operation - Must be "emit"
• event_type - Type of workflow event (see Event Types below)
• payload - Event-specific data object

Optional:

• workflow_id - Unique workflow identifier (auto-generated if not provided)
• work_id - Work item ID for workflow_id generation

Event Types:

Example Request:

{
  "operation": "emit",
  "event_type": "step_complete",
  "workflow_id": "workflow-199-20251202T150000Z",
  "payload": {
    "phase": "build",
    "step": {
      "name": "loader-validate",
      "skill": "corthion-validator",
      "status": "success",
      "duration_ms": 12500
    },
    "artifacts": []
  }
}

Step 1: Validate Request

Validate the incoming request:

1. Verify operation is "emit"
2. Verify event_type is one of the supported types
3. Verify payload is present

If validation fails, return error with details.

Step 2: Generate or Validate Workflow ID

If workflow_id not provided:

WORK_ID="${work_id:-unknown}"
TIMESTAMP=$(date -u +"%Y%m%dT%H%M%SZ")
WORKFLOW_ID="workflow-${WORK_ID}-${TIMESTAMP}"

Store in workflow state for subsequent events in same workflow.

Step 3: Build Event Object

Construct the full event with automatic fields:

{
  "event_type": "{event_type}",
  "workflow_id": "{workflow_id}",
  "timestamp": "{ISO 8601 UTC}",
  "project": "{detected from git or PWD}",
  "environment": "${FRACTARY_ENV:-development}",
  "payload": {payload}
}

Step 4: Detect Project and Environment

# Project detection
PROJECT=$(basename "$(git rev-parse --show-toplevel 2>/dev/null || pwd)")

# Environment detection
ENVIRONMENT="${FRACTARY_ENV:-development}"

Step 5: Write Event to Logs

Invoke log-writer skill with:

• log_type: "workflow"
• data: The constructed event object

Or use script directly:

scripts/emit-event.sh \
  --event-type "{event_type}" \
  --workflow-id "{workflow_id}" \
  --payload '{payload_json}'

The log-writer handles:

• Writing to .fractary/logs/workflow/
• S3 push (if configured in logs config)
• Retention policy

S3 Push Architecture:

The emit-event.sh script writes events to local storage only. S3 push is handled separately:

1. Via log-writer skill (recommended): Invoke log-writer instead of direct script call. The log-writer checks config and pushes to S3 if enabled.

2. Via background sync: Configure a background process or hook to sync .fractary/logs/workflow/ to S3 periodically.

3. Via CI/CD: Push logs to S3 as part of workflow completion in CI pipeline.

Configuration for S3 push (in .fractary/plugins/logs/config.json):

{
  "types": {
    "workflow": {
      "cloud_storage": {
        "enabled": true,
        "provider": "s3",
        "bucket": "${ORG}.logs.${PROJECT}"
      }
    }
  }
}

See plugins/logs/types/workflow/standards.md for complete S3 configuration options.

Step 6: Return Result

Return confirmation:

{
  "status": "success",
  "event_type": "{event_type}",
  "workflow_id": "{workflow_id}",
  "timestamp": "{timestamp}",
  "log_path": "{path to written log}"
}

Success:

Event emitted: {event_type}
Workflow: {workflow_id}
Timestamp: {timestamp}

Structured Response:

{
  "status": "success",
  "event_type": "step_complete",
  "workflow_id": "workflow-199-20251202T150000Z",
  "timestamp": "2025-12-02T15:10:00Z"
}

Invalid event type:

Error: Invalid event_type: '{event_type}'
Valid types: workflow_start, phase_start, step_start, step_complete,
             artifact_create, phase_complete, workflow_complete

Missing payload:

Error: payload is required

Log write failure:

Error: Failed to write event to logs
Details: {error}
Suggestion: Check logs plugin configuration

For Manager Skills

Add event emission at orchestration points in your manager skill:

<WORKFLOW>
## Initialize Logging

WORKFLOW_ID="workflow-${work_id}-$(date -u +%Y%m%dT%H%M%SZ)"

Skill("fractary-logs:workflow-event-emitter", {
  "operation": "emit",
  "event_type": "workflow_start",
  "workflow_id": WORKFLOW_ID,
  "payload": {
    "context": {
      "work_item_id": work_id,
      "entity_id": entity_id,
      "action": requested_action
    }
  }
})

## After Each Step

Skill("fractary-logs:workflow-event-emitter", {
  "operation": "emit",
  "event_type": "step_complete",
  "workflow_id": WORKFLOW_ID,
  "payload": {
    "phase": current_phase,
    "step": {
      "name": step_name,
      "status": "success",
      "duration_ms": step_duration
    },
    "artifacts": created_artifacts
  }
})

## At Workflow End

Skill("fractary-logs:workflow-event-emitter", {
  "operation": "emit",
  "event_type": "workflow_complete",
  "workflow_id": WORKFLOW_ID,
  "payload": {
    "status": "success",
    "duration_ms": total_duration,
    "summary": {
      "steps_executed": steps.length,
      "artifacts_created": artifacts.length
    }
  }
})
</WORKFLOW>

Minimum Viable Integration

For 80% of the value with minimal effort, emit just 3 events:

1. workflow_start - At initialization
2. artifact_create - When important outputs created
3. workflow_complete - At end with summary

S3 Configuration

Configure in .fractary/plugins/logs/config.json:

{
  "types": {
    "workflow": {
      "local_retention_days": 7,
      "cloud_storage": {
        "enabled": true,
        "provider": "s3",
        "bucket": "${ORG}.logs.${PROJECT}",
        "prefix": "workflow/{year}/{month}/"
      }
    }
  }
}

Reference

See WORK-00199 for full specification.