Files
gh-gtmagents-gtm-agents-plu…/commands/build-model.md
2025-11-29 18:29:43 +08:00

49 lines
3.0 KiB
Markdown
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
---
name: build-model
description: Generates a modeling plan detailing transforms, tests, and deployment schedule for analytics use cases.
usage: /analytics-pipeline-orchestration:build-model --use_case "pipeline velocity" --stack "dbt" --refresh daily
---
# Command: build-model
## Inputs
- **use_case** name of metric or dashboard relying on the model.
- **stack** modeling tool (dbt, LookML, Metrics Layer, SQL Runner, Python jobs).
- **refresh** cadence (hourly, daily, weekly) or cron.
- **dependencies** optional upstream tables or APIs.
- **tests** optional list of validations to enforce.
### GTM Agents Pattern & Plan Checklist
> Mirrors GTM Agents orchestrator blueprint @puerto/plugins/orchestrator/README.md#112-325.
- **Pattern selection**: Modeling often runs **pipeline** (spec → blueprint → testing → deployment → docs). If testing + deployment prep can parallelize, log a **diamond** segment with merge gate.
- **Plan schema**: Save `.claude/plans/plan-<timestamp>.json` with objective, data lineage, task IDs, parallel groups, dependency matrix, error handling, and success metrics (freshness %, defect ceiling, SLA adherence).
- **Tool hooks**: Reference `docs/gtm-essentials.md` stack—Serena for repo diffs/dbt patches, Context7 for platform docs, Sequential Thinking for review cadences, Playwright for UI validations tied to modeled data.
- **Guardrails**: Default retry limit = 2 for failed tests/deployments; escalation path = Analytics Modeling Lead → Data Engineering Lead → RevOps.
- **Review**: Run `docs/usage-guide.md#orchestration-best-practices-puerto-parity` before execution to confirm agents, dependencies, deliverables.
## Workflow
1. **Spec Alignment** review event/tracking plan, KPI definitions, stakeholders.
2. **Model Blueprint** outline staging, intermediate, mart layers, join keys, surrogate IDs.
3. **Testing Strategy** define schema, freshness, unique, accepted value, and custom tests.
4. **Deployment Plan** schedule jobs, resource configs, backfill strategy, rollback steps.
5. **Documentation & Handoff** update dbt docs / catalog, change log, owner assignments.
## Outputs
- Modeling spec (diagram, SQL pseudocode, dependencies).
- Test plan + configuration snippets.
- Deployment checklist with monitoring hooks and rollback instructions.
- Plan JSON entry stored/updated in `.claude/plans` for audit trail.
## Agent/Skill Invocations
- `analytics-modeling-lead` architects model + tests.
- `quality-gates` skill ensures validation coverage.
- `instrumentation` skill confirms data contracts stay intact.
## GTM Agents Safeguards
- **Fallback agents**: document substitutes (e.g., BI Publisher covering modeling reviews) when specialists unavailable.
- **Escalation triggers**: if freshness, defect, or SLA guardrails breach twice within 24h, escalate to Data + RevOps leadership per GTM Agents runbook and consider rollback.
- **Plan maintenance**: update plan JSON whenever dependencies, owners, or deployment cadence changes, keeping audit alignment with GTM Agents standards.
---