v2 draft outline
Outline only. Bullets list what this page will eventually cover.
Module 3: Make It Good
Sizing: 5 days (Week 3); second half of converting a bad repo to a good one
Outcome
- Close the gaps from your Module 2 plan
- Prove the gates actually work
What you'll do
- Execute the plan from Module 2 on Chaos Emerald Hotels (
liatrio-labs/chaos-emerald-hotels), highest risk first - Add what is missing: instruction file, hooks, CI, tests, scanners, dependency automation, deploy
- Fix what is wired in but ignored
- Prove each gate works by making it fail, then pass
- Run changes through the new factory as ordinary work
Closing the gaps
AGENTS.mdand written-down standards- Pre-commit hooks mirrored in CI
- Tests with a coverage floor
- Scanners at the right gate; the good repo is the reference, so OSV-Scanner goes in pre-commit as it does there
- Dependency automation
- Deploy and smoke test
Prove the gates work
- A gate you have never seen fail is a gate you cannot trust
- Trigger each gate with a deliberate problem
- Record what it caught and how it reported it
- Check that CI output is readable by an agent and a human
Bug fixes and small features
- Just work passing through the factory, not the point of the week
- Use them to exercise the gates, not as the deliverable
Gate
- You build and tune the gates for Chaos, using the good repo as the reference
- Placement is not yours to choose yet: that comes in the Capstone
- Decide where your own factory vs. hacktory line sits and defend it
Anti-patterns
- Closing gaps without proving the gates fire
- Hooks that fail silently where tooling is not installed
- Scanners running with findings ignored
- Heavy scans on every commit
- Green gates and an unread PR summary
Demo
- Process, not fixes: the gap list, what you closed, the gate you broke on purpose and what it said
From v1
- Replaces the old DevOps and Platform week
- Pipeline authoring becomes mirrored hooks and CI gates
- Terraform and IaC are not part of this module (parked on the overview)
TODO
- TODO: whether a learner may deviate from the reference OSV-Scanner placement in Chaos, or only in the Capstone (current decision: Capstone only)
- TODO: how "prove the gates work" is assessed (a checklist, a reviewer, or a demo)
- TODO: whether the Vercel deploy is prebuilt for Chaos or learners wire it
- TODO: who defines the smoke-test confidence bar (provided, or learner-defined as in the earlier draft)
- TODO: tooling must run in the learner setup, cloud agents included