Sprint & Flagship project tiers (subscription credits, tier flows)

Why / Context

PRD §6.2 defines three project SKUs segmented by review depth — Sprint (1–2 wks, self/AI review), Capstone (4–8 wks, shipped at MVP), Flagship (2–4 months, fully mentor-guided, weekly touchpoints, live defense). Selling all three lets the market self-segment on willingness to pay while routing scarce mentor hours to the tiers that need them. Mentor tier includes 1 Sprint/month; Career includes 1 Flagship/year (currently gated by the career-flagship-credit FeatureFlag).

Schema already in place — NO migration required

• Project.tier (SPRINT|CAPSTONE|FLAGSHIP), defenseRequired, mentorHoursBudget, durationWeeksMin/Max
• SubscriptionCredit(creditType: SPRINT_PROJECT|FLAGSHIP_PROJECT, periodKey, granted/usedCount) + ProjectInstance.subscriptionCreditId / source=SUBSCRIPTION_CREDIT
• ProjectReview.kind=AI_FIRST_PASS (Sprint's only review path)
• Price.mentorLevel for Flagship seniority pricing
• FeatureFlags seeded: sprint-tier, flagship-tier, career-flagship-credit
• PlatformConfig: projects.mentorAttributableBpsByTier, projects.reviewSlaHoursByTier

Implementation plan

1. Studio authoring: allow creators to select tier at project creation; tier-specific validation (Sprint: no review checkpoints, self-review checklist required in brief; Flagship: weekly milestone cadence + defenseRequired=true).
2. Sprint delivery flow: purchase → instance starts immediately (no PENDING_KICKOFF), structured self-review checklist UI, automatic ProjectReview(kind=AI_FIRST_PASS) on submission (depends on issue #5), auto-completion when checklist + AI pass.
3. Flagship delivery flow: weekly touchpoint scheduling (LiveSession(purpose: MENTOR_CHECKPOINT) recurrence helper), mentor capacity check (Flagship counts 3× against MentorProfile.maxActiveInstances), mandatory DefenseSession before PASSED (depends on issue #3).
4. Credit grant job: extend the existing subscription-credit scheduled function to grant SPRINT_PROJECT monthly (Mentor/Career) and FLAGSHIP_PROJECT yearly (Career, only when career-flagship-credit enabled).
5. Credit redemption at checkout: zero-price OrderItem path (source=SUBSCRIPTION_CREDIT), atomic usedCount increment with optimistic concurrency.
6. Catalog: tier filter chips, tier badges on cards, tier explainer page (pricing anchored per PRD §8.3 bands).
7. Mentor payout math: verify mentorAttributableBpsByTier flows for the new tiers in the ledger unit tests.
8. Flip sprint-tier / flagship-tier flags; seed 2–3 Sprint projects and 1 Flagship with a real partner brief (see PRD §6.2 "real-company briefs").

Acceptance criteria

• Sprint: buy → submit → AI first-pass → auto-complete, zero mentor hours consumed
• Mentor-tier subscriber redeems monthly Sprint credit; second redemption in same period is rejected
• Flagship: weekly checkpoints scheduled; PASS blocked until defense verdict recorded
• Career annual Flagship credit grants only when career-flagship-credit flag is on
• Ledger MENTOR_FEE entries reflect tier-specific attributable bps

Dependencies / sequencing

• MVP M10 (Capstone engine) is the foundation
• Issue #5 (AI first-pass) blocks Sprint auto-review; issue #3 (defense engine) blocks Flagship completion gate

Start with the existing subscription-credit scheduled function, the M10 Capstone engine, and the schema and FeatureFlags listed in the issue. Trace checkout redemption, tier delivery flows, and the ledger unit tests, while checking blockers #5 and #3. Done means the listed Sprint, Flagship, credit, catalog, and tier-specific payout acceptance criteria pass.