Skip to content

PRD Dependency Graph

Last Updated: February 2026

This document maps dependencies between PRDs so that prioritization decisions account for blocking relationships.


Dependency Matrix

graph TD
    PRD12[PRD 12: Billing] --> PRD2[PRD 2: Marketplace]
    PRD12 --> PRD3[PRD 3: Progression]
    PRD12 --> PRD4[PRD 4: Voxel World]
    PRD12 --> PRD6[PRD 6: Onboarding]
    PRD12 --> PRD10[PRD 10: Proactive]

    PRD3 --> PRD4
    PRD3 --> PRD10
    PRD3 --> PRD11[PRD 11: Persona Studio]

    PRD7[PRD 7: Memory] --> PRD9[PRD 9: Gateway API]
    PRD8[PRD 8: Safety] --> PRD9
    PRD11 --> PRD9

    PRD1[PRD 1: MiniApp Engine] --> PRD2
    PRD5[PRD 5: Group Chat] --> PRD6

    style PRD12 fill:#4CAF50,color:#fff
    style PRD1 fill:#4CAF50,color:#fff
    style PRD5 fill:#4CAF50,color:#fff
    style PRD7 fill:#4CAF50,color:#fff
    style PRD8 fill:#FFC107,color:#000
    style PRD6 fill:#FFC107,color:#000
    style PRD2 fill:#FFC107,color:#000
    style PRD10 fill:#FFC107,color:#000
    style PRD3 fill:#FF9800,color:#fff
    style PRD11 fill:#FF5722,color:#fff
    style PRD9 fill:#FF5722,color:#fff
    style PRD4 fill:#FF5722,color:#fff

Legend: Green = mostly shipped, Yellow = partially shipped, Orange = minimally shipped, Red = not started.


Dependency Details

PRD 12 (Billing) — Foundation Layer

Depended on by: PRDs 2, 3, 4, 6, 10 Blocking items for dependents: - PRD 2 needs subscription gating for marketplace installs (shipped) - PRD 3 needs subscription tiers to gate slot counts (not shipped — binary on/off only) - PRD 4 needs one-time purchase support for premium themes (not shipped — subscriptions only) - PRD 6 needs payment step in onboarding (shipped, but annual plan not in upgrade flow) - PRD 10 needs subscription check for proactive eligibility (shipped)

Status: Core billing is shipped. Gaps are slot mechanics, one-time purchases, and annual plan in upgrade flow.


PRD 3 (Progression) — The Critical Bottleneck

Depends on: PRD 12 (subscription tiers) Depended on by: PRDs 4, 10, 11 Blocking items for dependents: - PRD 4 needs levels/tiers to visualize in the voxel world (not shipped — only 4-stage relationship exists) - PRD 10 references "Gold-tier unlocks proactive behaviors" (not shipped — no tiers exist) - PRD 11 references persona modifiers driven by progression (not shipped)

Status: Only the 4-stage relationship system exists. The 10-level progression, slot mechanics, superpower tiers, and loadout system are entirely unbuilt. This is the single biggest bottleneck for downstream PRDs.

Recommendation: PRDs 4, 10, and 11 should define temporary stand-ins that work without PRD 3's full level system (e.g., use relationship stage, message count, or installed superpower count as proxies).


PRD 1 (MiniApp Engine) → PRD 2 (Marketplace)

Blocking items: - Marketplace depends on the superpower definition schema, install mechanics, and session runtime — all shipped. - Forward-compatibility (schema versioning, operation deprecation) is needed before marketplace scale.

Status: Non-blocking. PRD 1 is sufficiently complete for PRD 2.


PRD 5 (Group Chat) → PRD 6 (Onboarding)

Blocking items: - Group exposure is the primary viral acquisition channel described in PRD 6. - The onboarding DM flow (first group message → DM to non-user) depends on group detection — shipped. - Attribution tracking (which group led to signup) is not shipped.

Status: Core dependency is met. Attribution gap doesn't block onboarding, but blocks measuring group-as-CAC.


PRDs 7, 8, 11 → PRD 9 (Gateway API)

Blocking items: - PRD 9 exposes memory, safety, and persona to third parties. These systems must be stable first. - Memory (PRD 7): Shipped and stable. Needs provider-agnostic interface definition. - Safety (PRD 8): Shipped but violation storage is a TODO. Must be complete before exposing to third parties. - Persona (PRD 11): Persona engine is stable, but Studio/authoring tools aren't built. Gateway can launch without Studio.

Status: PRD 9 could proceed with memory and safety as-is, but violation storage and provider-agnostic interfaces should be completed first.


Based on dependencies and current state:

Priority PRD Rationale
1 PRD 12 gaps (annual upgrade, slot decision) Foundation for everything. Small effort, high leverage.
2 PRD 8 gaps (violation storage, safety regression tests) Required before scaling or exposing APIs. Safety debt.
3 PRD 3 Phase 0 (document current 4-stage system formally) Unblocks PRDs 4, 10, 11 to define fallback behaviors.
4 PRD 10 (enable proactive scheduler) Highest user-facing impact. Already built, just disabled.
5 PRD 2 gaps (ratings, ranking, creator dashboard) Marketplace polish. Engine works.
6 PRD 6 gaps (session persistence, resume, attribution) Onboarding reliability.
7 PRD 7 cleanup (provider-agnostic spec, namespace migration) Technical hygiene.
8 PRD 1 (forward-compat, schema versioning) Platform maturity.
9 PRD 5 (viral tracking, group memory policy) Measurement gaps.
10 PRD 9 (Gateway API v0) Requires all above to be stable.
11 PRD 11 (Persona Studio) Requires frontend. Lower urgency.
12 PRD 4 (Voxel World) Requires PRD 3 + frontend. Longest lead time.