Playbooks

Playbooks are standard operating procedures (SOPs) for IPM. Use calls.md for payload shape and required fields.

Surface availability (explicit)

• API Gateway: Available.
• Direct Lambda: Not offered.
• CLI: Available (g3n ipm ..., API Gateway).
• MCP: Available.

Playbook: Webhook subscription + replay

Goal: Register a webhook and replay deliveries for recovery.

Why this sequence:

• Webhooks must be registered and verified before use.
• Replay is the recovery mechanism for missed deliveries.

Preconditions

• Target webhook endpoint or queue is reachable and owned by the org.
• A valid org session/API key.

SOP (happy path)

1. Create webhook (POST /webhook/create).
   • Reason: registers destination + event filters.
2. Get/list webhooks (POST /webhook/get, /webhook/list).
   • Reason: verify status and configuration.
3. Set status (POST /webhook/status/set) to activate or pause.
   • Reason: operational control for delivery.
4. Replay (POST /webhook/replay) for a time window.
   • Reason: recover missed events from the store.
5. Query event catalog (POST /event/catalog/list) to confirm event coverage.
   • Reason: ensure the subscription covers the desired events.

Outputs

• Active webhook with verified configuration and replay capability.

Failure modes / remediation

• not-found: verify webhook ID or filters.
• invalid-state: ensure webhook is active before replay.

Cross-service relationships

• All services: publish events consumed via IPM.

Playbook: CDC feed + bulk export

Goal: Consume CDC events and run bulk export jobs.

Why this sequence:

• CDC provides near-real-time change feeds.
• Bulk export provides backfill and large snapshots.

Preconditions

• Export contracts defined for required event streams.

SOP (happy path)

1. List export contracts (POST /export/contract/list).
   • Reason: determine supported streams and schemas.
2. Read CDC (POST /cdc/list) with filters and cursors.
   • Reason: stream incremental changes.
3. Create bulk export (POST /bulk/export/create).
   • Reason: start a snapshot job.
4. Continue export (POST /bulk/export/continue) until complete.
   • Reason: paged export flow.
5. Check job status (POST /bulk/job/get, /bulk/job/list).
   • Reason: confirm completion and diagnostics.
6. Run bulk import (POST /bulk/import/create → upload → POST /bulk/import/commit).
   • Reason: ingest large data sets when needed.

Outputs

• CDC feed consumption and completed bulk export/import jobs.

Failure modes / remediation

• throttled: slow down and retry with backoff.
• invalid-state: confirm job status before continuing.

Cross-service relationships

• Accounting/SCM/PVM/etc.: IPM provides integration surfaces for their events.