TL-004 · Rules-engine generator
Tailored Process Checklist Generator
Most QA process checklists assume gold-standard access. Yours probably isn't. Toggle the access state your team actually has and the checklist on the right rebuilds live. Adding compensating items where you have gaps and dropping items you can't act on.
Your access state
Check what your team has today.
The checklist on the right rebuilds live as you toggle.
Tailored process checklist · 10 items
base 3 · conditional 0 · compensating 6 · modified 1
Code & access
- BaseMap every repo your team needs to read. Note which can be PR'd to and which can't.
- CompensatingResilient selector strategy: ARIA roles, text content, structural patterns. Avoid CSS class selectors.
- CompensatingTest ID request workflow: a documented path for getting test IDs added by the dev team.
Environment
- CompensatingWithout ephemeral envs, design tests to be tolerant of shared state. And keep them small.
- CompensatingDocument the environment-provisioning conversation as a Phase-0 blocker for leadership.
Observability
- CompensatingWithout app log access, capture HAR files + DOM snapshots on every failure. AI debugging will be slower.
AI tooling
- ModifiedAI is available but cannot execute. Limit to test-generation + review use cases until shell access is granted.
Org permissions
- CompensatingNo engineering sponsor. Surface this as the #1 adoption blocker. Most teams that skip this stall by week 4.
Pipeline
- BaseManifest-based decomposition: every task is broken into small subtasks the orchestrator tracks.
- BaseVerification owned by the orchestrator, not self-reported by the AI agent.
The full version checks 45 booleans and applies ~120 rules. Available inside the QA Leverage Review.
The legend
How to read the output.
- Base: applies regardless of your access state.
- Conditional: only when you have specific access. E.g. when your team can write to the app repo.
- Compensating: added when an access item is missing. These are workarounds for the constraint.
- Modified: a base item rewritten because some access is partial.
The full version inside the QA Leverage Review checks 45 booleans and applies ~120 rules. Including bucket classification and a section-by-section heat map.