What shipped
Built a focused local MVP for the Data Center Docket Scan Service Kit. The build reuses the prior Data Center Docket Watch package structure in this new build directory and adds a deterministic service-kit command that turns outputs/docket-watch.json into a sellable service package.
New runnable commands:
``bash python -m data_center_docket_watch service-kit --path /tmp/data-center-docket-watch-demo python -m data_center_docket_watch validate-service-kit --path /tmp/data-center-docket-watch-demo bash verify.sh ``
The service-kit command writes:
service-kit/offer-one-pager.mdservice-kit/client-intake-form.mdservice-kit/source-gathering-checklist.mdservice-kit/scope-and-disclaimer.mdservice-kit/proposal-and-sow.mdservice-kit/delivery-runbook.mdservice-kit/packet-qa-checklist.mdservice-kit/walkthrough-agenda.mdservice-kit/follow-up-email.mdservice-kit/monthly-watch-playbook.mdservice-kit/service-kit.json
Architecture
- Kept the stack Python 3.11+ and standard library only.
- Copied the prior build into this build directory instead of modifying the parent build in place.
- Added
data_center_docket_watch/service_kit.pyas a separate packaging module. - Made
outputs/docket-watch.jsonthe service-kit source of truth. - Added
validate_service_kit()plus avalidate-service-kitCLI command for deterministic acceptance checks. - Kept Markdown rendering as plain deterministic functions, matching the parent renderer style.
- Wrote JSON with sorted keys and stable indentation through the existing
write_json()helper.
Trimmed scope
- No hosted app, database, accounts, auth, billing, CRM, or dashboard.
- No live scraping of legislation, agenda, planning, utility, tax incentive, GIS, or meeting systems.
- No browser automation.
- No email sending or public-comment submission.
- No LLM calls and no Claude, Anthropic, OpenRouter, OpenAI, or other model-provider calls.
- Monthly watch is a manual playbook and upsell, not an automated monitoring engine.
Limitations
- The service kit depends on a previously generated
outputs/docket-watch.json; it does not auto-run the parent packet generator. - Source gathering is manual and checklist-driven.
- The seeded Nashville-like sample is demo data, not an official live record.
- The generated service documents are starting templates and still require human confirmation of current official dates, rules, documents, and votes.
Verification
bash verify.sh passes from the build root. It runs unit tests, recreates /tmp/data-center-docket-watch-demo, runs the parent sample/generate/validate flow, runs service-kit generation and validation, parses JSON outputs, and checks the required service-kit files and phrases.
Suggested next steps
- Try the service kit on one real client-provided docket workspace and tune copy based on delivery friction.
- Add optional client name and sender fields for follow-up email personalization.
- Add a lightweight changelog file for monthly watch refreshes if the recurring offer starts selling.
- Add more sample workspaces for different jurisdiction patterns only after the single-docket service proves useful.