FEATURE: [Feature name, e.g. "Orders – balance payment"] BRANCH: [Bubble branch the feature is built on] BRIEF: [Optional – paste or link the feature brief] PLATFORM: [Only if a specific platform is explicitly required. Otherwise leave blank – a single platform is assumed] GOAL Produce every distinct, reachable user flow that creates, changes or deletes data in this feature, for the QA Coverage Map. Work from the app's source only – do not open the app in a browser. Map data first: start from the data types the feature writes, find every workflow that writes them, keep only the ones a user can reach, then expand each into its flows. Flows that write no data (navigation, filters, search, sort, visibility) are out of scope. DEFINITIONS Reachable: a workflow a user can trigger by clicking from a page they normally land on, or a backend/scheduled workflow or webhook started by such a workflow or by an external service the app uses (e.g. a payment provider webhook). NOT reachable (list under "Not mapped" instead): - Workflows or actions that are disabled - Workflows whose trigger element is hidden by default with no condition or workflow that ever shows it - Workflows on reusable elements not placed on any page - Workflows only triggered from a screen reached by typing a URL or URL parameter by hand, with nothing in the app that generates that link - Backend/API workflows that nothing reachable schedules or calls Definition of a flow vs a variation: A flow is defined by what the user does: the clicks, the steps, their order, and the choices the user makes during the flow (e.g. habit type, applying a promo or not). If any of these differ, it is a separate flow. A variation is the same flow with an identical action sequence, run under a different context set outside the flow: user type or role, account state, admin or background settings, payment provider, or device/viewport. Variations go in the Variant columns of the parent flow's row, not in new rows. Test: identical Act, different Arrange → variation. Anything different in Act → new flow. Edge cases: - Same clicks with different entered data that changes the outcome (e.g. wrong password) → separate flow, because the result differs. - Incidental actions that don't change the outcome (e.g. show/hide password) → not flows. STEP 1 – List the data types List every data type this feature creates, changes or deletes (e.g. GP_Order, PayIntent, GP_Room). Include related data types its workflows also write (e.g. tickets created at checkout, a counter on the event). STEP 2 – List every workflow that writes them Search the whole app, not just the feature's pages, for every workflow with an action that creates, changes or deletes a record of those data types: page workflows, reusable element workflows, custom events, backend workflows, scheduled workflows, recurring workflows, and API workflows/webhooks. For each, note the data types and fields its actions write. STEP 3 – Keep only reachable workflows Trace each workflow back to what starts it: - Page or reusable workflow → its trigger element. Check the element is visible and clickable for at least one role: default hidden state, conditions that show/hide it or make it clickable, and workflows that show/hide it. Then trace back how the user reaches that element (page, popup, group) until you reach a page users land on. - Custom event → the workflows that trigger it; reachable if any of them are. - Backend, scheduled or API workflow → what schedules or calls it; reachable if that is reachable or is an external service the app uses. - Check the workflow and the writing actions are enabled. Anything unreachable goes to "Not mapped" with the reason. STEP 4 – Expand each reachable workflow into flows For each reachable workflow, create a separate flow for each of: - Each different reachable trigger: another button, popup or page that runs the same workflow. - Each "only when" condition on the workflow event or its writing actions that changes what gets written (e.g. create vs update, deposit vs full payment). - Each data-state gate: limit reached, record already exists, empty list, "Terminate this workflow", error or alert branches. The blocked path is its own flow. - Each user choice or entered value that changes what gets written (e.g. habit type, quantity 2, applying a promo). Put outside context (role, account state, admin/background setting, payment provider, device/viewport) in the Variant columns of the matching flow, not in new rows. When another role controls a setting that changes what gets written (e.g. Admin sets payment terms), changing that setting is its own flow for that role if it writes data; the affected flow carries the setting as a Variant. For combinations, include only pairs of settings that interact (one affects the other, or both affect the same total/record). List skipped combinations under "Skipped combinations". One flow per row, even when one user journey runs several workflows (e.g. checkout creates the order, pay intent and tickets) – that is ONE flow if the user does it in one go. Do NOT add: test-setup rows, cleanup rows, config-only rows, incidental actions, unreachable paths, or flows that write no data. Mark anything that looks like a bug or dead end from the source (e.g. a condition that can never be true, a field never set) with "BUG:" in Expected Outcome. STEP 5 – Follow the chain For each flow, follow every workflow it triggers, schedules or calls (custom events, backend and scheduled workflows, webhooks) and collect all the data they write. All of it goes in that flow's Data Validation. STEP 6 – Release checks If the feature adds or changes a field on an existing data type, add rows for: - Records created before the change (do they need a backfill or migration?) - Other workflows or screens that read or write the same field Category = "Release checks". Act may be empty; describe the check in Description and list the fields in Data Validation. STEP 7 – Make every flow checkable - Fixed example inputs: every typed or chosen value in Act is a literal in brackets, the same on every run (e.g. ACT-HABIT-NAME-INPUT(Deep work)). No placeholders, random values or "any". - Dedicated test user: Preconditions name the test account and its starting data (e.g. "Test user waterboy+habits@test.com with no habits"). Use test email addresses only. - Data Validation lists the data types and fields the flow's whole chain creates, changes or deletes. Only list them – no expected values, no calculations, no logic. STEP 8 – Output Return ONE csv code block, with this exact header and column order, one row per flow: Status,Platform,Category,Sub Category,Flow name,Preconditions (Arrange),Description,Action Sequence (ordered Action IDs) (Act),Expected Outcome/Goal (Assert),Data Validation,Priority,Automation (Now/Later/Manual),Role 1,Role 2,Role 3,Role 4,Role 5,Role 6,Logged Out User,Variant 1,Variant 2,Variant 3,Variant 4,Verification Column rules: - Status: leave empty. (Testers use only: Pass, Fail, Not tested, Blocked.) - Platform: the PLATFORM above only if explicitly given; otherwise leave empty. - Category / Sub Category: feature area / step (e.g. "Orders" / "Balance payment"). - Flow name: UPPER-KEBAB-CASE describing the path only (e.g. BALANCE-PAYMENT-DECLINED). No platform, app, feature or sub-category prefix. Unique across the file. - Preconditions (Arrange): the exact settings, field values, role and data state needed, written as Setting=Value (e.g. "Order Status=Partially Paid; logged in as Guest Consumer"). Never reference other rows. - Description: one sentence describing the journey. - Action Sequence (Act): ONLY action IDs, comma-separated, no prose, no numbering. Form: ACT---, where SCREEN is the page, popup or section (e.g. ACT-LOGIN-EMAIL-INPUT(user@test.com), ACT-LOGIN-SUBMIT-CLICK). No platform or app prefix. Put the chosen or entered value in brackets where relevant; no commas inside brackets. Reuse the same ID for the same action across rows. List the full sequence; never reference other rows. - Expected Outcome: the visible result for the tester only (messages shown, what's visible or blocked, where the user lands). No data fields. No line breaks. - Data Validation: the data types and fields to check, in this exact format so it can be parsed: DataType(action): field1; field2; field3 | DataType(action): field1; field2 - DataType: the Bubble data type name exactly as in the app (e.g. GP_Order, PayIntent). - action: one of created, changed, deleted. - Fields: Bubble field names exactly as in the app, separated by "; ". For deleted, write no fields: "GP_Room(deleted)". - Separate data types with " | ". No other text, no values, no line breaks. - Example: "GP_Order(changed): Order Status; Total Due Now; SecondaryPartialPayintents | PayIntent(created): AmountPay; PayIntentStatus" - For blocked flows, list the data types that must NOT change, with action "changed" (the tester checks they didn't). - Priority: High if it touches payments, record creation or deletion, access control, or another user's data; Medium for common variations; Low for rare paths. - Automation: "Now" for High priority deterministic paths, "Later" otherwise, "Manual" if it needs visual judgement or an external system (e.g. email inbox, payment provider hosted form). - Role 1–6: one column per user type in the app (e.g. Role 1 = End User, Role 2 = Admin); state the mapping under "Role key". "Yes" for every role that can perform this flow; otherwise empty. Unused role columns stay empty. - Logged Out User: "Yes" if a logged-out visitor performs this flow; otherwise empty. - Variant 1–4: outside context only (role, account state, Admin/background setting, payment provider, device/viewport) as Setting=Value with its expected difference, one per column (empty if none). If more than 4, list the rest under "Skipped combinations". - Verification: "Inferred-From-Source" for every row. - Quote any field containing commas, quotes or line breaks. No line breaks inside fields. - Before returning, confirm every row has exactly 24 columns. After the csv block, add these short sections (NOT part of the CSV): 1. "Role key" – which user type each Role column represents. 2. "Workflow coverage" – a table: Workflow (page/type) | Data types written | Flow names that run it. Every reachable workflow from Step 3 must have at least one flow name. 3. "Cross-role dependencies" – a table: Setting | Set by (role, page) | Values | Changes flows for (roles) | Flow names affected. 4. "Brief coverage" – only if a BRIEF is given: each requirement | flow names that cover it. Flag requirements with no rows and anywhere the source differs from the brief. 5. "Not mapped" – unreachable workflows, with why each is unreachable. 6. "Skipped combinations" – combinations or extra variants left out, and why. 7. "Uncertain" – anything you couldn't determine from the source (e.g. behaviour inside a plugin or external API), and which rows depend on it. Do not change the app. Do not write or apply tests.