UAT guide for teams & human users
Chat-first UAT: uat list → start uat <id> → pass N / fail N → uat done. Analysts: qa status, qa failures, approve sync CODE (Dev→QA gate, no auto-rsync). Self-train quizzes + sign-off + PROD backup-before-promote. Open /businessite/uat/help and /businessite/uat/train.
Chat commands
uat help · uat list · start uat <id> · pass N · fail N · uat done · qa status · qa failures · qa run · approve sync CODE · guide uat · /businessite/uat · /businessite/uat/help · /businessite/uat/train
Why UAT is critical (and rewarding)
Every solid test reduces operating cost (less firefighting), unlocks growth (safe features for customers), and lets people earn continuously (support shifts, 3% trading, business digitisation). UAT is not bureaucracy — it is how BusinesSite stays trustworthy.
Golden rule: nothing reaches PROD without documented steps, critical Pass sessions, required sign-offs, and an automatic PROD backup before QA→PROD sync. Dev→QA is gate + notify only — approve in chat/dashboard; host sync is deliberate.
Chat is the main UAT interface
Prefer /businessite/chat (web or mobile hub) for feature testing:
uat list— features ready / in testingstart uat <id or name>— open a session and show stepspass 1·fail 2 short reason— record each stepuat done— submit session (critical fails create Analyst tasks)uat session— show current progress
Web runner remains at /businessite/uat if you prefer the form UI.
Analysts: QA agent loop on Dev
Continuous cron runs smoke/UAT HTTP checks on Dev. Results live under BusinesSite → UAT Testing → QA Agents and in chat:
qa status/uat results— latest runsqa failures·qa failures data(layer filter)qa run— managers trigger smoke suite nowqa sync list·approve sync CODE·reject sync CODE
Approving sync writes a gate JSON under the UAT gate folder; it does not auto-rsync to QA.
Docs: module docs/QA_AGENT_LOOP.md.
Who participates (5 teams + humans)
Analysts lead as test managers and lead testers. They coach:
- Data — records, reports, migrations, cost numbers
- System — instances, gate runner, backup/restore path, QA agent suites
- User — registration, chat, profile, internal login, chat-first UAT
- Security — portal vs internal, owner claims docs, ACL
- App — web UX, mobile hubs TA5/RC1/RC2/TA13, media, QR link
Human users test their own profile and businesses/services (eShop, co-hosted like Colourcraft Auto).
Analyst playbook
- Watch QA Agents → Runs (or
qa status) for Dev regressions. - Create a UAT Release (backend: BusinesSite → UAT Testing → Releases).
- Add Features + steps (instructions, test data, expected, critical flag, modules).
- Release for UAT — broadcasts chat/push; testers use
uat list. - Coach sessions; watch Timeline history (immutable).
- When Dev is green: Request Dev→QA sync →
approve sync(gate only) → host runs sync. - On QA: collect sign-offs; then Promote to PROD (backup then sync).
Tester playbook (all layers)
- Chat:
uat help·uat list·start uat …· pass/fail ·uat done. - Or open QA / web /businessite/uat.
- Self-train: /businessite/uat/train (quizzes).
- Critical Fail → Analyst task; re-test after fix.
Human profile & business testing
- Profile: sector, role, mother tongue, photo, internal user works for chat.
- Business/eShop: shop name (not personal name), mall vs domain, products, payments.
- Owner claim:
join business Colourcraft Auto as owner(docs + admin).
Self-train & quizzes
Complete quizzes on /businessite/uat/train. Pass ≥70% to earn badges.
Say guide uat or uat help in chat anytime.
stop or cancel.
Say continue to stay.