Build from a fixed brief
Keep the starting conditions visible.
- Named tool and time budget
- Published functional requirements
- No silent manual repair before the audit
Resource 15 · Open engineering series
The monthly teardown
A repeatable public teardown format for testing AI-built applications against the same nine-check production standard.
The working guide
Follow the evidence across the whole system.
Each edition will publish the brief, build constraints, original result, evidence-backed findings, remediation decisions, and final retest. No invented case study and no hidden scoring changes.
Run the same standard
Make scores comparable.
- All nine checks and thresholds
- Evidence gaps shown separately
- Critical-path behavior tested directly
Publish the failure modes
Show what actually broke.
- Finding, severity, and evidence
- Business or user consequence
- Why the original implementation allowed it
Repair and retest
Close the loop.
- Chosen remediation and alternatives rejected
- Before-and-after evidence
- Remaining limits and operational ownership
Evidence to keep
Leave a record another person can verify.
- 01Original brief and repository snapshot
- 02Initial score and evidence table
- 03Change log with remediation rationale
- 04Final score and unresolved risks
Choose the next useful step
Use the guide to define the first accountable decision.
We can review the evidence with your team and turn the gaps into a scoped plan without taking ownership away from you.