Implementation Tracker (Phase 1 MVP)

As of: 2026-09-22 · Spec: PRODUCT_CONTEXT.md

Legend: 🟢 Done · 🟡 Partial · 🔴 Not Started · ⚪ Dropped · 🔵 Open/TBD

---

Summary Statistics

Metric	Count	Share
Total requirements tracked	58	100% (C01–C18 + FR01–FR19 + AC01–AC21)
🟢 Done	2	3%
🟡 Partial	28	48%
🔴 Not Started	13	22%
⚪ Dropped	9	16%
🔵 Open / TBD	6	10%

---

Product Boundary Requirements (C01–C18)

ID	Requirement	Status	Evidence / Notes	Source
C01	Standalone SaaS; separate deployment, credentials, data boundaries	🟡 Partial	Sophia monorepo + own Mongo/S3/Stripe/Fargate path; shared cleverhire-bot and no production CleverHire adapter. Independence gate AC16 not qualified.	§3.1(1), §7.1, AC16
C02	Public self-service signup without routine Highbrow activation	🟢 Done	POST /v1/auth/signup, Google signup, workspace creation (auth.routes.ts, auth.service.ts). ONBOARDING still says sales-led — stale.	§3.1(2), §7.1 C02, §23.1
C03	Stripe USD checkout; public product/pricing/help/policy pages	🟡 Partial	Stripe Checkout + webhook (credit.routes.ts, stripe-webhook.routes.ts). No public marketing/pricing/help/policy pages or sample report in apps/web.	§3.1(3), §12.3
C04	Prepaid minute credits; max-duration reserve; charge actual usage	🟡 Partial	Lots, ledger, reserve/settle/release (credit.service.ts, LedgerCreditsProvider). Trial grant optional (TRIAL_CREDITS_*) — conflicts with “no free live credits” when enabled.	§3.1(9), §7.1 C04, §8.1
C05	Exactly two roles: admin and member; creator starts admin	🟢 Done	EUserRole.ADMIN / MEMBER in @sophia/shared; first signup user is admin.	§7.1 C05, §23.2
C06	Members: own interviews only; no private billing	🟡 Partial	Credit routes admin-only; billing UI gated. Interview list/detail not filtered by createdBy for members (interview.routes.ts comment vs interview.service.ts).	§7.1 C06, §4.2
C07	Admins: all interviews, team, billing, per-scheduler usage	🟡 Partial	Team, credits summary/history/lots, admin delete interview. No usage/spend CSV export API.	§7.1 C07, FR14
C08	One workspace/account; last-admin protection; immediate role effects	🟡 Partial	countActiveAdmins + team demotion guards (team.service.ts). Multi-workspace per account / immutable creator not fully verified.	§7.1 C08, §6.2
C09	Cognito; Google + email/password; no enterprise SSO	🟡 Partial	Firebase Auth + Google (auth.firebase.ts) — functional overlap, spec mismatch vs Cognito.	§7.1 C09, §23, T-002
C10	Shared templates + bulk scheduling	🔵 Open / ⚪	Templates ⚪ dropped (10 Sep). Bulk CSV/XLSX fate unresolved — no bulk UI/API.	§7.1 C10, D13, D26, Q14
C11	Google/Microsoft calendar; slots; self-booking	⚪ Dropped	Day-window + ICS email replaces calendar integration.	§7.1 C11, D14
C12	English 1:1 AI interview; 5/10/15/20/30 min (default 20)	🟡 Partial	Duration on positions/interviews; candidate live via Daily + bot session. Full browser matrix / enforcement not qualified.	§7.1 C12, contracts
C13	Sofia first; CleverHire consumes service later	🔵 Open	Strategy tension with AGENTS.md module rule (§23.3). No production adapter.	§7.1 C13, G5
C14	Broad ATS compatibility later phase	🔵 Open	Zod contracts only; no adapter tests shipped.	§7.1 C14, FR18
C15	First-time user/candidate usability as release gate	🔴 Not Started	No recorded 5-participant usability qual (AC19).	§7.1 C15, §17.3
C16	Delivery ownership, four-week plan, QA week	🔵 Open	Process/meta; milestones need re-baseline (§18.6).	§7.1 C16
C17	Competitive pricing; viable measured cost	🔴 Not Started	Prices in spec; no blended $0.15/min measurement in repo.	§7.1 C17, §17.4
C18	Phase 1 phone/person/eye + credibility; dual camera Phase 2	🔴 Not Started	proctoring.credibility schema + PATCH from bot; no FR19-grade detectors in Sophia web. Dual camera correctly out of scope.	§7.1 C18, FR19, §23.11

---

Functional Requirements (FR01–FR19)

ID	Requirement	Status	Evidence / Notes	Source
FR01	Signup/join: public signup, workspace, first admin; invitee joins intended workspace	🟡 Partial	Signup + Google + verify flows; team invite accept (AcceptInvitePage.tsx). Invite email dev-only (DevInvitationProvider).	§7.2 FR01, §6.1–6.2
FR02	Accounts: reset, verification resend, logout; no unsafe identity merge	🟡 Partial	Firebase client flows + API logout/revoke; linking rules not fully tested.	§7.2 FR02
FR03	Membership conflicts: no auto-transfer; no domain auto-join	🟡 Partial	Single org per user model; edge cases not qualified.	§7.2 FR03
FR04	Calendar consent separation	⚪ Dropped	Calendar integration removed.	§7.2 FR04
FR05	Shared recruiter templates	⚪ Dropped	Positions replace template library.	§7.2 FR05, D13
FR06	Bulk CSV/XLSX scheduling	🔵 Open	Not implemented; fate undecided (Q14).	§7.2 FR06, D26
FR07	Draft caps, resend limits, immutability after send	🔵 Open	Resend rate limit in API; draft/staging caps not enforced.	§7.2 FR07, §5.16
FR08	Calendar integration	⚪ Dropped	—	§7.4
FR09	Appointments / self-booking	⚪ Dropped	Day-window model instead.	§7.4
FR10	Candidate runtime: private pages, consent, screen share, recovery, FR19 detectors	🟡 Partial	Join, lobby, device checks incl. screen share (useDeviceChecks.ts), Daily live. No phone/person/true gaze; not full browser matrix.	§7.5 FR10
FR11	Calendar/appointments (continued)	⚪ Dropped	—	§7.4
FR12	Results: honest skills, transcript, recording, integrity, notes, PDF/export	🟡 Partial	Report generation with assessed + insufficient_evidence (report-generation.service.ts). No report UI, PDF, or evidence export routes.	§7.5 FR12, T-015
FR13	Credits/commercial packs; integer credit-seconds; charge rules	🟡 Partial	Packs via Stripe; ledger types; settle charges floor(seconds/60). Chargeback/deficit not built.	§7.6 FR13
FR14	Admin billing dashboard, alerts, member funding messages	🟡 Partial	Billing page + ledger; funding alerts / member-only messaging incomplete.	§7.6 FR14
FR15	Email + in-product notifications	🔴 Not Started	No notification consumer; T-017 open.	§7.7 FR15
FR16	Retention, export windows, purge, holds	🟡 Partial	retentionExpiresAt on interview; settlement sets default 60d. No purge sweeper, export pipeline, or holds.	§7.7 FR16, T-018
FR17	Ops: capacity limits, monitoring, operator console, support roster	🔴 Not Started	No operator routes; no 10/3 concurrency enforcement; no on-call artifacts in repo.	§7.7 FR17
FR18	Extraction: versioned contracts, adapter tests, reuse bot pipeline	🟡 Partial	Bot context/start/complete routes; shared Zod contracts. No adapter contract test suite; bot still CleverHire-coupled in prod.	§7.8 FR18
FR19	Phase 1 proctoring: phone/person/eye + credibility (amended)	🔴 Not Started	Proctoring blob + credibility fields on model; no qualifying detector pipeline in Sophia.	§7.9 FR19, D16–D17

---

Acceptance Criteria (AC01–AC21)

ID	Requirement (summary)	Status	Evidence / Notes	Source
AC01	Public signup → workspace → first admin; resumable abandoned signup	🟡 Partial	Signup + org + admin role implemented; resumable signup not qualified.	§6.1, AC01/AC02 refs
AC02	Invitee joins intended workspace/role; safe identity linking	🟡 Partial	Invite accept flow; email delivery stub; takeover edge cases not tested.	§6.2, §7.2
AC03	Two workspaces; tamper across APIs/files — zero leakage; member billing isolation	🔴 Not Started	Org-scoped repos + rules; no automated AC03 suite; member interview scope gap (C06).	§16.1, §17.3
AC04	One-workspace / invite conflict handling	🟡 Partial	Team service + invite expiry fields; full WAL22 scenarios not run.	§6.2
AC05	Positions, candidates, schedule prepared inputs (résumé, skills-fit)	🟡 Partial	CRUD + skills validation + resume upload; readability gate partial.	§6.4–6.6, FR07 survivors
AC06	Bulk import scheduling (100 rows)	⚪ Dropped	Pending Q14 / QA re-baseline.	§3.3, Q13
AC07	Google Calendar integration	⚪ Dropped	—	§3.3, Q13
AC08	Candidate self-booking / slots	⚪ Dropped	Day-window replaces.	§3.3, Q13
AC09	Preflight failures → no live charge	🟡 Partial	Billing starts at live session; preflight not wired to meter end-to-end qual.	§6.7, WAL08
AC10	Live session, disconnect/recovery, terminal state + transcript preserved	🟡 Partial	Bot disconnect/rejoin routes; disconnected status on model; sweeper for abandoned not built.	§6.8–6.9, T-018
AC11	Wallet / ledger invariants (WAL01–WAL24 subset)	🟡 Partial	Atomic reserve/settle, Stripe idempotency; ≥1k sequence qual not run.	§15.1
AC12	Stripe purchase, webhook grant once, pending/failed grant nothing	🟡 Partial	stripe-webhook.routes.ts, duplicate handling; full WAL01–02 not qualified.	§6.3
AC13	Day-window scheduling + invitation (ICS)	🟡 Partial	expiryDays 7/14, window on interview; ICS + production email not sent.	§6.6, D14
AC14	Report quality: evidence, insufficient-evidence, no fabricated skills	🟡 Partial	LLM prompt enforces assessed; human ≥20-review gate not run.	§6.11, §17.3
AC15	Retention notices, export before purge, deletion/hold	🔴 Not Started	Retention timestamp only; no purge/export/hold jobs.	§6.14, FR16
AC16	Independence: CleverHire blocked; full paid journey on Sophia only	🟡 Partial	Data/API on Sophia; bot dependency + no formal AC16 test bundle.	§5.17, §10.10
AC17	Measured blended cost ≤ $0.15/paid minute (release gate)	🔴 Not Started	No cost qualification artifact.	§3.3, §17.4
AC18	Platform failure injection: alerts ≤60s, replay, backup/runbooks	🔴 Not Started	SQS DLQ infra; no fault-injection qual or runbooks in repo.	§15.3
AC19	Usability: 5 first-time users, ≥4/5 core tasks without coaching	🔴 Not Started	—	§17.3, Q31
AC20	Load: 10 concurrent (3/workspace), soak, 100-session day mix	🔴 Not Started	Bot default 3 cluster-wide; no qual harness.	§17.2
AC21	Proctoring qualification (PRO01–10, held-out set, false escalation)	🔴 Not Started	FR19 not implemented to gate level.	§15.2, §17.3, Q15

---

Decisions Register (D1–D28)

ID	Decision	Spec status	Implemented?	Conflicts / notes
D1	Standalone public SaaS	FINAL	🟡 Partial	Own stack yes; shared bot + no AC16 sign-off.
D2	Reuse shared Python bot	FINAL	🟡 Partial	Bot routes in API; production E2E incomplete.
D3	Single shared DB + organizationId	FINAL	🟢 Done	Models/repos follow data-access rules.
D4	Ledger + cached org credits + reconciler	PROVISIONAL	🟡 Partial	Cache + ledger yes; reconciler job missing (T-018).
D5	Reserve at schedule, settle at end, settledAt claim	FINAL	🟢 Done	settlement.service.ts, credit reserve in schedule.
D6	Charge actual seconds, not reserved max	FINAL	🟡 Partial	Settlement uses elapsed seconds; WAL matrix not qualified.
D7	Frozen snapshot at schedule	FINAL	🟢 Done	snapshot on interview; schedule copies resume text.
D8	joinTokenHash on interview	PROVISIONAL	🟢 Done	Hash stored; password + session token flow.
D9	One candidates collection	FINAL	🟢 Done	Candidate module + per-org email unique.
D10	assessed + insufficient_evidence	FINAL	🟢 Done	Report generation + Zod contracts.
D11	deletedAt soft delete	PROVISIONAL	🟢 Done	Used on core entities.
D12	Integer credit-seconds	FINAL	🟡 Partial	Ledger integers; display rounding rules partial in UI.
D13	Templates dropped	FINAL	⚪ N/A	Positions only.
D14	Day-window + ICS; calendar dropped	FINAL	🟡 Partial	Window in code; ICS email not production.
D15	Google + email Phase 1; no SSO	FINAL	🟡 Partial	Firebase not Cognito.
D16	Hybrid eye + AI-likeness → credibility	FINAL	🔴 Not Started	Schema only.
D17	Phone/person/eye remain Phase 1	FINAL	🔴 Not Started	—
D18	Dual camera Phase 2	FINAL	⚪ N/A	Correctly out of scope.
D19	positionId required	FINAL	🟢 Done	Interview model requires position.
D20	Separate interviewReports	FINAL	🟢 Done	Report model + generation consumer.
D21	Transcript in Mongo for now	PROVISIONAL	🟢 Done	Inline on report document.
D22	Free-text tags ~20 max	PROVISIONAL	🟡 Partial	Candidate tags; convention not enforced in schema.
D23	Global-unique user email	PROVISIONAL	🟢 Done	Unique index on user email.
D24	disconnected + failureReason enum	PROVISIONAL	🟢 Done	On interview model/status enum.
D25	Skills + aiContext on position	PROVISIONAL	🟢 Done	Position create + snapshot skills.
D26	Bulk scheduling undecided	TBD	🔵 Open	No implementation.
D27	Role naming admin/member	TBD	🟢 Done	Code uses admin/member (Q21 effectively resolved in code).
D28	Rejoin state embedded on interview	TBD	🟢 Done	runtime subdocument on interview.

---

Open Questions (Q1–Q32)

ID	Question	Status in code / product
Q1	Global vs per-org user email	Resolved in code: global unique (D23).
Q2	Store positions.description for bot	Resolved: snapshot + bot context use JD text.
Q3	Transcript Mongo vs S3	Resolved for now: Mongo on report (D21).
Q4	interviewSessions collection	Resolved in code: embedded runtime (D28).
Q5	credits.consumed rounding	Resolved by spec: floor seconds at settlement.
Q6	Tags free-text vs managed	Open: free-text in code; degeneration accepted (D22).
Q7	seniority field value	Open: kept; maps to experience level in practice.
Q8	Rejoin embedded vs collection	Resolved: embedded (D28).
Q9	Pin botType provider	Open: field exists, default openai.
Q10	ICS shape + reminder timing for windows	Open: invitation subdoc; no production ICS.
Q11	Eye-tracking on-device vs cloud	Open: D16 not implemented.
Q12	Spec v0.5 lock date	Open: blocks formal QA sign-off.
Q13	QA contract re-baseline	Open: AC06–08 still in PDF appendix.
Q14	Bulk CSV fate	Open: not built.
Q15	AC21/PRO gates after credibility redesign	Open: engineering must re-validate.
Q16	Credibility threshold + compute location	Open: no production logic.
Q17	Candidate reschedule/cancel on window model	Open: likely N/A; confirm with Kevin.
Q18	No-show / late entry semantics	Open: expired window = no-show in working theory.
Q19	Default window length options	Partial: org default 7; UI allows 7/14.
Q20	Candidate password/challenge storage	Partial: join password on interview; security review open.
Q21	admin/member vs owner/member	Resolved in code as admin/member; docs lag (§23.2).
Q22	Self-serve vs sales-led	Resolved by product: self-serve; ONBOARDING stale.
Q23	Ownership matrix	Open: follow PDF Appendix B; update ONBOARDING.
Q24	Module inside CleverHire vs consumed service	Open: packages/ neutrality rule binds.
Q25	3 → 10 concurrent bots	Open / blocker: not implemented.
Q26	Redis role	Open: not used in Sophia code paths.
Q27	Which LLM paths run for reports	Open: OpenAI in report consumer; duplicate analysis not mapped.
Q28	Contract versioning for ATS	Open: /v1 only.
Q29	Admin 2FA / operator hardening	Open: not discussed in spec as required.
Q30	QA $500 / infra $500 approval	Open: process/budget.
Q31	Proctoring + usability participant recruitment	Open: AC19/AC21 blockers.
Q32	Who runs fresh held-out qual after detector fix	Open: ownership unassigned.