dv-cook

name: dv-cook description: End-to-end project execution wrapper for Data Visualization Kit. Use when the user types `$dv-cook` or wants to continue a project after `$dv-plan` with the context, dataset, and goals already locked.

Treat $dv-cook as the canonical post-intake execution wrapper.

Follow ./.codex/workflows/cook-workflow.md as the source of truth.
Use this workflow after $dv-plan has already created the project workspace and captured the full intake.
Run the project in order:
- $dv-data-preparation
- $dv-data-visualize
- local deploy — start the selected dashboard tool locally so the user can test it (see Local Deploy Rules below)
- project validation / test — check files AND verify the local dashboard is reachable
- $dv-document-management
- $dv-publish
Keep all project outputs inside projects/<slug>/docs/.
Re-enter specialist workflows directly only when one stage needs a targeted refresh.
Use the project-scoped harness artifacts created by $dv-plan:
- docs/validation.md for proof matrix status
- docs/stories/ for story-sized work
- docs/decisions/ for inherited workflow decisions
- docs/assets/harness/traces.jsonl for workflow trace history
- docs/drift-audit.md for stale or contradictory project-state checks

The final dashboard MUST be a live, interactive UI — not a static HTML export.

Evidence.dev: run npm run dev inside projects/<slug>/evidence/ — the dev server is the live interactive UI
Metabase: start via Docker Compose — the running Metabase instance is the live interactive UI
Grafana: start via Docker Compose — the running Grafana instance is the live interactive UI

Never deliver a flat HTML file or screenshot as the final dashboard. The user must be able to interact with it in a browser.

After $dv-data-visualize completes, determine which tool was selected from projects/<slug>/docs/visualization.md and start it locally:

Tool	Local start command	Default URL
Evidence	`cd projects/<slug>/evidence && npm install && npm run dev`	http://localhost:3000
Metabase	`docker compose up -d` from `projects/<slug>/` (requires `docker-compose.yml`)	http://localhost:3000
Grafana	`docker compose up -d` from `projects/<slug>/` (requires `docker-compose.yml`)	http://localhost:3000

Confirm the server starts without errors before proceeding to validation.
Tell the user the local URL where the dashboard is running.
If the tool requires Docker and Docker is not available, report it as a blocker.

During the validation stage, perform tool-aware checks:

File checks — verify required docs and assets exist (run test skill or project validator)
Tool-aware deploy check — based on the tool in docs/visualization.md:
- Evidence: confirm npm run dev starts and http://localhost:3000 is reachable
- Metabase: confirm Docker container is running and Metabase UI is reachable
- Grafana: confirm Docker container is running and Grafana UI is reachable
Harness checks - inspect docs/validation.md, docs/assets/harness/traces.jsonl, and docs/drift-audit.md for stale or contradictory state.
Report the live URL to the user so they can open and test the dashboard directly.