Empire Teas CRM — PRD Signals
Sidecar to
project-prd.md. Carries live signals only — open and partial open questions that the engagement still needs to settle.
What lives here
- Open questions raised during PRD authoring, the three engagement gates (
/forge-prd-check,/forge-arch-probe,/forge-decompose), or mid-engagement when a new ambiguity surfaces. - Only
⏳ openand◐ partialrows. Once an OQ is answered (✅), it moves toproject-prd-history.md## Resolved Open Questions, with the answer folded into the PRD body itself. See.claude/rules/prd.mdfor the resolution procedure.
What does NOT live here
- The live contract — problem, domain, scope, NFRs, success criteria — lives in
project-prd.md. - Risks — risks live in
project-prd.md## Risks. They are part of the engagement contract; OQs are not. - Resolved questions and PRD revisions — both live in
project-prd-history.md.
Why the split exists
Mixing live signals and audit trail into the PRD body bloats every read of the contract. Selective loading on the consumer side (e.g. spec-reviewer filtering signals by feature ID via the Blocks column) is only possible when signals are in their own file. See .claude/rules/prd.md for the full rationale and the shape-guard hook contract.
Status Legend
- ⏳ open — no answer yet
- ◐ partial — answer in progress; some aspects settled, others outstanding. Also used for
ASSUMEDrows — a working assumption is in place (recorded in the PRD body and the project proposal) but the client has not yet confirmed it; the row stays live until they do.
(Answered ✅ rows live in project-prd-history.md — not here.)
Assumptions pending confirmation (2026-08-02)
The project proposal was written against a set of working assumptions for questions the client had not yet answered; it has since been issued and commercially accepted (2026-08-16). The assumptions are the ◐ partial rows tagged ASSUMED below — recorded as assumptions, not confirmed requirements, in both the PRD body and the proposal’s Assumptions section — and each is to be confirmed at contract signature / kickoff (target ~Fri 28 Aug / Tue 1 Sep 2026). If a client answer differs from the assumption, it may change scope, timeline, or cost via the change process. The highest-impact assumption still carrying risk is OQ-2 (manual ERP handoff for V1); OQ-15 (external cloud AI) was resolved — accepted 2026-07-28, arrangements signed off with the accepted proposal (Rev 12) — and its processing-location half was reopened on 2026-09-30 (Rev 48); that half is open again in the table below.
Open Questions
Ordering: the original OQ-1…30 rows run ascending; OQs raised later append at the bottom, newest last.
The
Sectioncolumn anchors each OQ to a PRD body section (e.g.§Domain Model,§Functional Surface > [Feature X]) so the answer’s eventual home is obvious.The
Blockscolumn lists feature IDs the OQ blocks — comma-separated, or—if none. Thespec-reviewersub-agent filters this table by feature ID during/forge-spec-review, so only OQs relevant to the spec under review are loaded into its context.Feature IDs are the sibling ids assigned at Gate 3 (
/forge-decomposeRun 1, 2026-09-17 —F5-c,F11-b, …; the PRD’s F-number is each sibling’s parent, and an un-suffixed id such asF12is a feature that was not cut). The live index isfeatures.md; dropped items (F27–F30, F35; Rev 66) keep their PRD ids; F37, a stretch candidate at Gate 3, was cut into F37-a and F37-b when committed to Phase 2 (Rev 57).
| # | Section | Topic | Question | Status | Resolution / Owner | Blocks |
|---|---|---|---|---|---|---|
| OQ-1 | §Scope & Boundaries | Pipeline mix | Approximately how many active private-label clients vs bulk buyers do you have? (Validates the two-pipeline design and its relative weighting.) Reclassified 2026-07-22 — count orgs/deals per pipeline in the Pipedrive audit rather than ask; more accurate than an estimate. 2026-10-06: no Pipedrive admin access or walkthrough is coming; Empire shared list exports of leads, organisations and people (13 and 28 Sep), which carry no deals. Answered from the Pipedrive deals export Empire is expected to share for Phase 2 (client/dependencies.md §1.3). |
⏳ open | Dinil (Team — Phase 2 deals export) | F2-a, F24 |
| OQ-2 | §Functional Surface > F19 / §Integration Points | ERP identity & API | Current ERP confirmed = IFS. V1 scope decision (2026-08-02): V1 ships the manual handoff only (per AD #10); automated IFS integration is a separately-priced follow-on once IFS API availability is confirmed — not in the V1 fixed scope. Residual to confirm before the follow-on: whether IFS exposes usable APIs / an export format. 2026-11-12 (the lead, from the 10 Nov progress meeting): Empire is replacing IFS with another ERP not yet chosen, so the residual is now which ERP replaces IFS and its integration details; until Empire names it and gives them, any automated link is out of scope and V1 gives only the handoff document (F19). | ◐ partial | Dinil | F19 |
| OQ-7 | §Domain Model | Tea grades in use | Which tea grades do you work with most (BOP, BOPF, OP, Dust, …)? Fixes the grade dropdown on deal/brief/quote fields. Reclassified 2026-07-22 — grades are visible in products/deal titles during the Pipedrive audit, and the standard Ceylon taxonomy is already researched (sri-lanka-tea-industry.md §5). 2026-10-06: no Pipedrive admin access or walkthrough is coming, and the list exports carry no deals; the weekly price workbook’s GRADE column already lists the grades of every standard priced that week (discovery/source-schemas/weekly-price-workbook-schema.md), and the Phase 2 deals export can confirm them. |
⏳ open | Dinil (Team — weekly price workbook) | F2-c, F7-a, F8-a |
| OQ-8 | §Domain Model | Volume units & pack format | What does a typical private-label order look like — volume and pack format (teabags/loose/tins/sachets)? Fixes the volume unit and packaging dropdowns. Reclassified 2026-07-22 — read from actual deal records in the Pipedrive audit. 2026-10-06: no Pipedrive admin access or walkthrough is coming; read from the Pipedrive deals export Empire is expected to share for Phase 2, if its deal titles or values carry volume and pack format, otherwise asked of Empire (client/dependencies.md §1.3). |
⏳ open | Dinil (Team — Phase 2 deals export) | F2-c, F7-a, F8-a |
| OQ-9 | §Functional Surface > F9 / §Functional Surface > F1 | Standard↔cert & buyer-cert mapping | Buyer-required certs are tracked in a spreadsheet (confirmed). Still open (2026-08-03 meeting): how the certifications a buying standard holds are stored is being clarified — action items: Kavindu to check with Kavi Amali; Praveen to provide the QC documentation. The F9 cert-mismatch check is gated on receiving this standard-cert data. (Reworded 2026-09-15, PRD Rev 27: the standard carries its held certifications — there is no product catalog.) Narrowed 2026-09-17 (team working assumption, relayed by Biseni at the F1-a gap check): a certification is held per buying standard — the per-batch/lot grain raised in the 3 Sep question set (Q36) was an internal concern, not a client signal, and is retired; F26-b builds the standard↔certification join per standard and F9 compares per standard. What stays open is the client half only: where Empire’s standard-held certification data lives today, as the seed / migration source for F26-b. The account-required side (the certification reference list and the account↔certification join) is settled and owned by F1-a, so F1-a is no longer blocked. F26-b released 2026-09-29 (developer waiver at its gap check): the join and its editing do not depend on where Empire keeps the data today; loading it is a seed / migration step once the source is known (F26-b spec, waiver B-2). |
⏳ open | Empire (Kavi Amali / QC) via Dinil | F9 |
| OQ-12 | §Functional Surface > F18 | PL stale threshold | How long does a typical private-label deal take from first inquiry to first order? Sets the private-label stale-deal alert threshold. Reclassified 2026-07-22 — compute from historical deal timestamps in the Pipedrive audit (stale-alert thresholds stay admin-configurable regardless). 2026-10-06: no Pipedrive admin access or walkthrough is coming; computed from the Pipedrive deals export Empire is expected to share for Phase 2 (client/dependencies.md §1.3). |
⏳ open | Dinil (Team — Phase 2 deals export) | F18-a |
| OQ-13 | §Functional Surface > F18 | Bulk stale threshold | How long does a typical bulk deal take from first inquiry to first order? Sets the (different) bulk stale-deal alert threshold. Reclassified 2026-07-22 — same method as OQ-12, computed from historical deal timestamps in the Pipedrive audit. 2026-10-06: no Pipedrive admin access or walkthrough is coming; computed from the Pipedrive deals export Empire is expected to share for Phase 2 (client/dependencies.md §1.3). |
⏳ open | Dinil (Team — Phase 2 deals export) | F18-a |
| OQ-16 | §Users and Access | AI review ownership | Each rep reviews their own AI outputs with manager oversight (confirmed 2026-07-22). ASSUMED (pending confirmation): the client’s “Auto or Manual” request governs how AI work is routed to a reviewer, not whether a human approves — human approval on every AI action remains mandatory (the locked governance invariant). F11-h released 2026-10-16 (developer waiver at its gap check): a person presses Send on every drafted email and nothing is drafted unasked, which meets the default whatever the routing answer; routing drafts to another reviewer would be new work on top of F11-h (F11-h spec, decision 1, W-1). F11-g released 2026-11-05 (developer waiver at its gap check): the AI ranking changes nothing on a lead and sends nothing — a person reads it and decides — so the routing answer does not change it (F11-g spec, decision 1, W-1). | ◐ partial | Dinil | — |
| OQ-17 | §Non-Functional / §Functional Surface > F12 | Inbound languages | What languages are inquiries typically received in (English, Arabic, Turkish, Russian)? Gates multilingual AI extraction (deferred). Reclassified 2026-07-22 — sample real inbound emails stored in Pipedrive during the audit (contingent on email sync being enabled, itself checked in the audit). 2026-10-06: no Pipedrive admin access or walkthrough is coming. The people export shows Pipedrive’s email sync was in use (email counts and last-sent/received dates are filled), but no export carries the email text, so this is to be asked of Empire directly when the lead raises the Pipedrive questions with them (owed.md). Note 2026-10-26: F12 leaves the Blocks cell — it is done, and multilingual extraction is deferred past V1 (PRD § Deferred). |
⏳ open | Tharinda → Empire Teas (ask later) | — |
| OQ-18 | §Scope & Boundaries (Deferred) / §Functional Surface > F13 | Document sources | Where are product specs, certifications, and SOPs stored (Drive / SharePoint / physical)? Prerequisite for the deferred RAG knowledge/compliance assistant. Deferred — not asked at kick-off (2026-07-22): only gates a deferred feature; ask when the RAG assistant is scheduled. | ⏳ open | Dinil | — |
| OQ-19 | §Integration Points / §Functional Surface > F19 | ERP payload | What customer/order data must move CRM→ERP on Won, and what (order/shipment/invoice status) should come back? Defines the handoff payload and any sync-back. Deferred — not asked at kick-off (2026-07-22): blocked by OQ-2 (which ERP is the actual handoff target); V1’s manual handoff content is already defined in the PRD. Ask per-ERP once the integration target is confirmed — possibly twice, if Empire integrates with the current ERP before the new one lands. | ⏳ open | Dinil | F19 |
| OQ-22 | §Non-Functional > Performance | User & license count | How many active users are on Pipedrive today (across which offices)? Sizes concurrency, licensing, and rollout/training scope. Reclassified 2026-07-22 — read the active-user count directly from the Pipedrive admin/export during the audit. 2026-10-06: no Pipedrive admin access or walkthrough is coming. The exports name 17–18 distinct record owners, which may include people who have left, so the active count is to be asked of Empire directly when the lead raises the Pipedrive questions with them (owed.md); the list of CRM users comes with client question Q1 (roles and the reporting line). |
⏳ open | Tharinda → Empire Teas (ask later) | F5-c |
| OQ-23 | §Scope & Boundaries (Out of Scope) | Pipedrive add-ons in use | Which paid Pipedrive add-ons does Empire use today — LeadBooster (chatbot/live chat), Campaigns (email marketing), Data Enrichment, any LinkedIn tool? Scope decision 2026-08-20: F27–F30 are out of V1 regardless of the answer; on 2026-10-11 they were dropped, with no later phase planned (PRD Rev 66). The answer decides no scope. It serves one purpose: early warning where reliance exists, so the gap against the “nothing lost on migration” promise is raised with the client at that time (as a change request) rather than surfacing at UAT. Ask it early for that reason. 2026-10-06: no Pipedrive admin access or walkthrough is coming, and the list exports cannot answer this; to be asked of Empire directly when the lead raises the Pipedrive questions with them (owed.md). |
⏳ open | Tharinda → Empire Teas (ask later) | — |
| OQ-25 | §Functional Surface > F20 / F25 | Templates & sequences in use | Do they use email templates and sequences today — how many, and what cadences? Sizes template migration and sequence design. Reclassified 2026-07-22 — inventory templates/sequences directly from the Pipedrive account during the audit. 2026-10-06: no Pipedrive admin access or walkthrough is coming, and the list exports cannot answer this; to be asked of Empire directly when the lead raises the Pipedrive questions with them (owed.md). Note 2026-10-26: F20-d leaves the Blocks cell — it is done, with this question waived for its spec (Rev 72). |
⏳ open | Tharinda → Empire Teas (ask later) | F25-a |
| OQ-26 | §Functional Surface > F4 / §Domain Model | Pipedrive audit inventory | What custom fields, saved reports, active automations, and connected integrations exist in Pipedrive, and which are business-critical? Plus migration data-quality inputs: how long has Pipedrive been in use, and are activities logged consistently across the team (does activity history warrant migration)? (Pipedrive audit Q12–Q16, Q42, Q45.) Ensures no relied-upon field/report/automation/integration is lost and the migration is scoped to real data quality. 2026-09-28 (PRD Rev 44): custom fields and activities are no longer migrated, so the answer no longer shapes the migration; F4-a and F4-c are removed from Blocks, and the question stays open for the reports, automations and integrations F17-a and F24 need. 2026-10-06: no Pipedrive admin access or walkthrough is coming, and the list exports cannot answer this; to be asked of Empire directly when the lead raises the Pipedrive questions with them (owed.md). |
⏳ open | Tharinda → Empire Teas (ask later) | F17-a, F24 |
| OQ-27 | §Functional Surface > F26 / F4 | Buying-standards list source | Where does the full buying-standards list live today — the ~35,000 issued numbers with their client, grade and specification (the weekly workbook carries only the ~1,450 priced that week) — and can we have a copy? Since PRD Rev 27 there is no product/SKU catalog to seed: the standards book is the F26 seed and the F4 migration source, and the private-label items must arrive as standards too. (Brand tracking is covered by OQ-20.) Reclassified 2026-07-22 — confirm the source during the Pipedrive audit; reworded 2026-09-15 (Rev 27) from “product/SKU list source”. F26-b released 2026-09-29 (developer waiver at its gap check): the standards book’s writes and search do not depend on where the list lives; loading it is F4-d’s (F26-b spec, waiver B-3). 2026-10-06: no Pipedrive admin access or walkthrough is coming, and the list lives outside Pipedrive; it is already asked as client question Q5. 2026-10-13 (F4-d spec, Decision 1): loading the list is not part of the Pipedrive migration — the list lives outside Pipedrive — so F4-d leaves the Blocks column; F23-b (Phase 2) offers a spreadsheet import, and how the list reaches the CRM in Phase 1 is an owed.md item for the lead. |
⏳ open | Empire Teas (client question list, Q5) | — |
| OQ-28 | §Functional Surface > F24 | Weighted forecast & rotting usage | Do you use weighted forecasting / stage probability and idle-deal (rotting) alerts today, and with what thresholds? Sets default stage probabilities and rotting days per stage per pipeline. Reclassified 2026-07-22 — visible in Pipedrive admin settings during the audit; overlaps OQ-26/OQ-30. 2026-10-06: no Pipedrive admin access or walkthrough is coming, and these are settings, not data, so no export carries them; to be asked of Empire directly when the lead raises the Pipedrive questions with them (owed.md). |
⏳ open | Tharinda → Empire Teas (ask later) | F24 |
| OQ-29 | §Scope & Boundaries (Out of Scope) | Sales goals usage | Do you use Pipedrive Goals (rep/team targets vs actuals) today, and with what metrics/periods? Scope decision 2026-08-20: F35 is out of V1 regardless of the answer; on 2026-10-11 it was dropped, with no later phase planned (PRD Rev 66). The answer decides no scope; it gives early warning if Empire actively relies on Goals today — raise any such gap with the client at that time, not at UAT. 2026-10-06: no Pipedrive admin access or walkthrough is coming, and the list exports cannot answer this; to be asked of Empire directly when the lead raises the Pipedrive questions with them (owed.md). |
⏳ open | Tharinda → Empire Teas (ask later) | — |
| OQ-30 | §Functional Surface > F24 / F2 | Current pipeline setup | How many pipelines exist in Pipedrive today, and what are the current stage names for each? The current stages are the starting point for the F24 seeded defaults — designing stages in a vacuum produces something unfamiliar the team won’t adopt. 2026-10-06: no Pipedrive admin access or walkthrough is coming; read from the pipeline and stage on each deal in the Pipedrive deals export Empire is expected to share for Phase 2 (client/dependencies.md §1.3). |
⏳ open | Dinil (Team — Phase 2 deals export) | F24, F2-a |
| OQ-49 | §Non-Functional > Deployment and Hosting | Two environments or three? | The issued proposal contradicts itself, so this needs a client answer rather than an internal edit. § 11 Deliverables promises Empire Teas “running development, staging, and production environments” at handover. But A.1 states “Environments: Two — development and production”, A.2.1 has development “double as the UAT environment”, A.2.2’s cost table carries only production and development columns, A.2.3 prices a separate staging environment as an unpurchased +$45/mo add-on, A.2.4 frames the three-environment configuration as the rejected ~$520/mo conventional option, and A.8’s total is captioned “both environments.” Five places in the appendix say two; one deliverables clause says three. Until settled, architecture.md and the PRD follow the costed appendix (two). If §11 governs, add the staging environment, its +$45/mo, and the extra promotion tier — and note it changes the CI/CD pipeline shape. Surfaced by the Gate 2 verification dry-run, 2026-08-06. Update 2026-08-18: delivery now runs each UAT round in a temporary environment stood up per round (or pre-go-live production with test-data cleanup) — dev is build-only; the two-permanent-environment costed basis is unchanged, and §11’s staging wording still needs the client’s written answer. Update 2026-09-27 (second harness review): SOW ET-002 as drafted (v1.2, client/aws-sow.md) states “The baseline comprises development and production environments” and excludes “additional staging, testing or regional environments”. If it is signed as drafted, that settles two-versus-three, but leaves one point to confirm in writing before the Phase 1 UAT window (2 Nov): whether the temporary UAT environment per acceptance round counts as such an “additional testing environment”. Closing this OQ is the lead’s call. Update 2026-09-29: SOW ET-002 is superseded — Empire Teas holds the AWS account and is billed by AWS directly (client/aws-sow.md header; proposal post-issue note of 29 September). The ET-002 question above therefore falls away: a temporary UAT environment is an ordinary cost on Empire Teas’ own bill, stated in client/dependencies.md §1.1, and no fixed-fee exclusion applies. §11’s “staging” wording still needs the client’s written answer, and the Phase 1 UAT window is now 30 Nov – 10 Dec (client/project-timeline.md). |
⏳ open | Dinil | — |
| OQ-41 | §Scope & Boundaries / §Reference-System Classification | Core Pipedrive reliance | At a high level, how does the team use Pipedrive day-to-day, and which features would they most miss / consider essential? Open-ended catch-all backing the “lose nothing on migration” promise. Decided 2026-07-27: raised live during the Pipedrive audit walkthrough rather than in writing — it gets a richer answer in conversation. 2026-09-28: F4-a removed from Blocks — the answer informs parity work generally, not how the import engine reads and maps the exports. 2026-10-06: no Pipedrive walkthrough is coming; to be asked in conversation at a regular meeting with Empire when the lead raises the Pipedrive questions with them (owed.md). |
⏳ open | Tharinda → Empire Teas (ask later, in conversation) | — |
| OQ-55 | §Functional Surface > F36 | Capture-app scope vs. what the POC already ships | Running forge-et-pwa-poc (read-only reference) shows it diverges from F36 in both directions. Already built, beyond F36’s definition: campaign-as-scoring-lens, bulk pile-splitting + camera-roll import, pre-show exhibitor CSV targeting + ranking, expo-intelligence campaigns, duplicate detection. Required by F36 but absent from the POC: badge QR scanning and offline-tolerant queued upload — verified absent (no QR/barcode library, no BarcodeDetector, no IndexedDB or background sync; the install banner’s “works offline” refers only to the Workbox app-shell precache, so a capture taken with no signal is currently lost). Working position: F36 stays exactly as written — the five extras are not V1 scope (post-V1 / descope-inventory material), and the two gaps are net-new build inside F36’s estimate. Needs the client’s confirmation rather than an internal edit, because the POC has been demoed: features they have seen working would be deliberately left out of V1. Also re-check the AD #20 sizing (~2.5–3.5 wk prototype-adjusted stream) against this — the frontend is more complete than that basis assumed, while QR + offline queueing are unbuilt. Empire answered in part on 2026-09-09 (PRD Rev 23): badge QR scanning is needed (encoding unknown — a badge photo is requested); duplicate detection is wanted as a scan-time existing-client check — the app tells the rep at once whether the organisation or person is already a customer and how often Empire deals with them, which needs a read lookup of accounts and contacts from the app (architecture Rev 33) and a “check pending” state offline; new: a same-day generic follow-up email to stand visitors (OQ-60 — dropped from V1 on 2026-10-11 with F36-f, PRD Rev 63); campaign planning is already present in the POC and is not a new ask; offline behaviour (Q23) was not discussed. The other extras (pile-splitting, camera-roll import, exhibitor CSV) stay out. F36 therefore grew — recorded against R-ARCH-001 (tracker note 2026-09-11). 2026-10-11 (PRD Rev 64): the offline queue gap is closed by dropping it — F36-e leaves V1, and F36-a keeps a capture whose upload fails and resends it when the app is open with a connection. 2026-11-12 (PRD Rev 96): the pre-show exhibitor targeting is in — the lead (Tharinda, reported by Nanndish) put planning before a show in Phase 1, built in F36-a (screens and data) and F36-c (research and ranking of every exhibitor); pile-splitting and choosing several photos stay out (Rev 95: one capture at a time); camera-roll import stays out. 2026-11-18 (PRD Rev 101): the campaign scoring lens is in, as a campaign’s optional focus (the lead’s decision, reported by Dinil; built in F36-c). What stays open here is the badge QR encoding (a badge photo). Until it arrives, F36-a’s spec reads the common QR formats — a vCard, a MECARD, a web link and plain text — and pins the badge’s real format when the photo comes (Nanndish’s waiver of 2026-11-12, recorded in the F36-a spec as D-13). F36-a leaves the Blocks column on 2026-11-12 under that waiver, as F11-c and F11-f left theirs; when the photo arrives, F36-a’s FR-C3 and AC-12 are revised. F36-b, F36-c and F36-d leave the Blocks column on 2026-11-13. Nanndish confirmed that his waiver extends to them (decision D-1 in each spec), because each works on whatever company, contact, email and website the badge or card gives: the badge’s real format changes only what F36-a’s FR-C3 reads. |
◐ partial | Tharinda (lead — client contact) | — |
| OQ-57 | §Domain Model / §Functional Surface > F1, F2 | Several parties on one lead (participants) | Empire (2026-09-09): an existing customer starts a new project and brings agencies in as stakeholders — several organisations and people, all on one email thread, connected to one lead, and Empire wants them grouped under it (the parties can be read from the email itself). Today a lead converts to one Account/Contact and a Deal is N:1 Account; the 3 Sep set’s Q35 default (flat organisations, agent = customer) is challenged. Proposed direction: one primary account per lead/deal plus a participants link (N:N contacts, each with a role — buyer / agency / agent), which also drives thread linking (Pipedrive parity: deal participants). Empire said it still “needs thinking about” — confirm the shape with them before the F1/F2 specs. Raised 2026-09-11 from the 9 Sep meeting (PRD Rev 23). F11-f released 2026-10-06 (developer waiver at its gap check): the lead page stores nothing about participants and keeps only an empty place for them, so a participants link added later changes neither (F11-f spec, decision 1, W-1). F1-d released 2026-10-08 (developer waiver at its gap check): the account page stores nothing about participants and has no place for them; a participants panel added later is a new place on the page (F1-d spec, decision 2, W-2). |
⏳ open | Dinil | F2-c |
| OQ-59 | §Functional Surface > F26 / F23 | Weekly price workbook ingestion | Empire’s weekly prices arrive as an Excel workbook by email (Weekly Auction Prices 35.xlsx, seen 2026-09-11): one tab per market, one row per standard — CLIENT · STD · TYPE · GRADE — one column per auction week; ~1,450 standards priced in a current week of ~35,000 numbers ever issued; hand-maintained (sparse CLIENT column carrying working notes, one shifted header, one tab without week columns, a few #REF!, text in price cells; the same standard number appears in more than one tab in ~30 cases, half with different recipes — drift and mis-keyed numbers, not a pricing rule). V1 takes it in as draft prices for the Tea Room to review and publish (PRD Rev 23). To settle with the Tea Room: the currency and unit of the numbers (rupees per kg? one tab converts at 327.5 LKR/USD, while OQ-44 has the published price FOB Colombo in USD) and how the USD figure is derived; who produces the sheet; upload into the CRM versus pick-up from the mailbox (automatic inbound triggering is post-V1 — default upload); a fixed template versus an importer that tolerates today’s shape; the duplicate-number rule (flag, never pick silently). Update 2026-09-13 (Empire, via Tharinda; the workbook’s structure is now extracted to .forge/discovery/source-schemas/weekly-price-workbook-schema.md): confirmed — standard numbers are unique to one customer (a customer holds many; the ~30 cross-tab repeats are data drift to flag, not a pricing rule), no per-region price (tabs are navigation groupings only), and the derivation is LKR → USD at a rate (≈ 327.5 LKR/USD) → + profit margin → final USD; the Tea Room can manually edit a weekly price per customer-standard and upload the workbook (default; automatic mailbox pick-up is post-V1); weekly history is persisted (AD #14) for trends. Update 2026-09-17 (team decision — PRD Rev 31): the derivation half is CLOSED. A weekly price is stored as a four-figure build-up — price in LKR → + overhead (LKR) → converted to USD → + margin → final USD price, of which only the final USD figure is required. All four sit on the append-only price row (AD #14); the data-model change is owed to Tharinda (AD #31). Still to settle with the Tea Room: whether the exchange rate and the overhead are one figure per week or one per standard — this decides whether they live on a per-week row (e.g. auction_sale) or as columns on every price, not whether the build-up is kept; the unit of the base numbers (rupees per kg?); who produces the sheet; and fixed-template vs. tolerant-importer. F26-a released 2026-09-25 (developer waiver at its gap check): the four build-up columns do not depend on the remaining half (forge-et-be/docs/data-model.md §2.2, §2.14 item 6), so F26-a is specced without it; F26-c settles the rate/overhead grain before it builds the importer (F26-a spec W-1, FR-D2). Update 2026-10-06 (team decision — Tharinda, recorded by Dinil at F26-c’s gap check; PRD Rev 53): two more halves CLOSED. (1) Rate and overhead are kept per price row: the LKR-per-USD rate the conversion used is a column on the price row, and the overhead is the difference between the two LKR figures already there — this holds whether Empire uses one rate a week or several, and the workbook already carries different rates on different tabs in one week (327.5 and 327). (2) No fixed template: the upload accepts the workbook as it is today and reports every row or cell it cannot read instead of guessing. The duplicate-number rule was already settled (flag, never pick). Still open, for the Tea Room: the unit of the LKR figures — F26-c builds on LKR per kg and shows the unit on screen, so a wrong assumption is visible — and who produces the sheet (anyone holding buying_standard.manage can upload, so the answer changes no build; it names who runs the upload at UAT). F26-c no longer blocked (2026-10-06): it builds on LKR per kg (F26-c spec D-1), and the two open points change only labels and the UAT contact, so F26-c leaves the Blocks cell. |
◐ partial | Tharinda (data-model owner) | F23-b |
| OQ-61 | §Domain Model / §Functional Surface > F26, F7 | Private-label standards — weekly-priced or quoted per deal? | PRD Rev 27 (2026-09-15) made the Buying Standard the sellable record for private label as well as bulk (the products catalog is removed; the prototype seeds private-label standards). OQ-37’s Rev 2 answer said only bulk pricing is price-sensitive and private label is quoted bespoke per deal. To settle with Empire / the Tea Room: does a private-label standard carry a weekly published price like a bulk standard, or is it priced on the quote — a standard with no weekly price row, for which F7’s “unpriced standard blocks quoting” must not fire? Does it carry a buying cost for the margin floor? Default until answered: private-label standards are quoted per deal — no weekly price row is expected for them, and the F7 gate treats a private-label standard as priceable on the quote. F26-f released 2026-10-10 (developer waiver at its gap check): the weekly round builds on the default — a private-label standard cannot be drafted or published in it — and an answer the other way removes that refusal by a spec revision (F26-f spec, waiver B-1, D-3). | ⏳ open | Tharinda (data-model owner) / Empire via Dinil | F7-d |
| OQ-63 | §Functional Surface > F1 / §Integration Points | Past orders on the account — no V1 feed | PRD F1 promises past orders sourced from the ERP (read-only, “via ERP”) on the account detail, and the §Domain Model carries an Order entity with the ERP as its source of record. Nothing feeds it in V1: OQ-2 settled the ERP boundary as a manual CRM→ERP handoff document (AD #10) with the ERP→CRM read (order / shipment / invoice status) explicitly deferred, so no order data reaches the CRM before the separately-priced IFS follow-on. Gate 3 decision (2026-09-17, team answer Q11-A): F1-d renders no past-orders panel in V1 and F1-a carries no order table; F13’s account briefing synthesises from the timeline and open items without order history. To put to Empire: (a) is a manual “last order” field on the account (date, standards, value — typed in by the rep or the Tea Room) worth having as a stop-gap until the ERP read exists, or (b) do they wait for the follow-on? Default if no answer: (b), nothing built — the panel appears with the ERP integration. Raised by /forge-decompose Run 1 (finding 1). F1-d released 2026-10-08 (developer waiver at its gap check): F1-d builds default (b) — no past-orders panel and no manual last order field; if Empire chooses (a), it is a revision of the F1-d spec (decision 1, W-1). F13 released 2026-10-21 (developer waiver at its gap check, Dinil): F13 builds default (b) — the account briefing reads no order history, and its reorder/upsell trigger becomes a re-engagement prompt on dormant accounts; if Empire chooses (a), the last order field joins the briefing’s input in a revision of the F13 spec (decision 1). |
⏳ open | Dinil (client-facing — via Tharinda) | — |
| OQ-66 | §Functional Surface > F1 / §Domain Model > Account | Are our account lists and certification list right? | Raised 2026-09-24 from F1-a spec Q-5. The certification list (FR-R1) and the account dropdown values — customer types, regions, packaging formats, Incoterms (FR-R2) — were written by Altrium from the prototype, not given by Empire (R-03). They ship as V1 defaults; each wrong entry costs one small release (OQ-65: the lists are code-declared in V1). To put to Empire, in the weekly update: are these lists right, and is the certification list meant to be the certifications a buyer requires (what the account field records) or the ones Empire holds (what FR-R1 seeds) — probably not the same set. Default if no answer: the lists stand, corrected when the Pipedrive export shows the real values. | ⏳ open | Dinil (client-facing — via Tharinda), in the weekly update | — |
| OQ-67 | §Functional Surface > F1 / F18 | How many days of silence make an account “dormant”? | Raised 2026-09-24 from F1-b Q-2 and F1-c’s open question. The account list shows a Dormant badge (PRD F1). F1-a wave 3 computes it from the last timeline entry against a threshold that is a backend setting, default 90 days (F1-a spec Rev 4, FR-A1), because Empire has not named the number. To put to Empire, in the weekly update: after how many days without any contact should an account count as dormant — and should it differ for bulk and private-label buyers? Default if no answer: 90 days for every account. Changing the number is a configuration change, not a code change; F1-c owns the final rule. F1-c released 2026-10-13 (developer waiver at its gap check): F1-c keeps F1-a’s rule — every timeline row counts, lead rows included — with the number as one setting, 90 days for every account; Empire’s answer, including a split by buyer type, is a configuration change or an F18-a addition (F1-c spec, decisions 1–2, FR-D1, FR-D2). Note 2026-10-26: F1-c leaves the Blocks cell — it is done, built on the waiver above. |
⏳ open | Dinil (client-facing — via Tharinda), in the weekly update | F18-a |
| OQ-68 | §Functional Surface > F20 | Should email sent from the CRM carry the rep’s signature? | Raised 2026-10-12 by the F20-c spec (Q-1). Mail the CRM sends goes through Microsoft Graph, which does not add the signature the rep set up in Outlook, and Graph offers no way to read it. Pipedrive lets each user keep a signature in its own settings. To put to Empire: do reps expect their signature on email sent from the CRM? If yes, the answer is a per-user signature setting in the CRM, added to every new email, reply and forward — a follow-up feature, not a change to F20-c. Default if no answer: no signature is added; the rep types one when wanted. The default is in force, so this question blocks no feature; F20-c ships without a signature. | ⏳ open | Tharinda (client contact) | — |