Incident Triage + Proof

This bundled starter workflow (wf-incident-triage-proof) models ingest → context enrichment → agent triage → a condition that routes high-severity work to a GitHub-backed escalation path, or automatic remediation on the other branch. Execution is verifiable: runs produce structured logs and a proof record linked to the run.

• GitHub / tool nodes are API-backed: wire integration_configs and credentials in your environment for live calls.
• Open the graph from Workflows — the definition is immutable; persists runs under your account.

See also Execution proofs and DAG execution.

Trigger shape (sample payload)

{ "sampleAlert": { "id": "INC-0001", "severity": "high", "service": "api" } }

Node-by-node

1. Ingest — webhook-style ingress.
2. Enrich — internalEcho → /api/workflow-echo.
3. Triage — agent stub (non-policy model defaults low severity; policy-style models can flip the gate).
4. Approval gate — severity routing; TRUE → GitHub integration gate.
5. GitHub integration? — TRUE → Escalate & Record (live GitHub issue via your saved PAT and repository: "__default__"); FALSE → internal stub.
6. Remediate — tool stub on the low-severity branch.
7. Slack integration? (after remediate) — TRUE → Slack: remediation notice (incoming webhook POST); FALSE → internal stub.

Run with sample inputs

POST /api/workflows/run with workflow_id: "wf-incident-triage-proof" and the JSON context above (or the builder button).

Expected logs

Same lifecycle events as other presets; no integration_missing on guarded GitHub or Slack nodes when the respective integration is not configured (stubs run instead).

Proof summary

See successfulTaskCount / failedTaskCount / integrationFailures on the proof row after the run completes.