# V7 Product Review — Mawu: The Open World & Creator Republic Status: independent product review — revised and expanded in a second meticulous pass on 2026-07-07 (§9, Additional considerations) and a third pass on 2026-07-07 covering monorepo integration, shared functionality, and TODOS/ roadmap alignment (§10) Reviewer: Claude (product-perspective deep review) Date: 2026-07-07 (first, second, and third passes) Scope: `V7/V7_features.md` + all `V7/features/*` pages, `V7_TODOS.md`, `V7_GAP_ANALYSIS.md`, `REMEDIATION_2026-06-12.md`, `ADVERSARIAL_EVAL_GATES.md`, all `V7/docs/*` policy papers, `release/launch-readiness-review.md`. Method: full deep-read, then critique against the shipped competitive set (FiveM/Cfx.re + the ESX/QBCore framework economy, Roblox + DevEx, Fortnite UEFN/Verse + engagement payouts, mod.io console pipeline, Nexus Collections, Steam Workshop, Garry's Mod/s&box, ArmA Reforger, Star Citizen server meshing) and the 2025–26 UGC-platform climate (Hytale's cancellation, the Roblox safety reckoning and age-assurance rollout, UK OSA/EU DSA/TAKE IT DOWN enforcement, GTA VI's imminence and the "Project ROME" rumor cycle). Market facts reflect knowledge through early 2026. --- ## 1. Executive summary V7 is the platform play of the V-series: not an authored game but the substrate for **community-run persistent worlds** ("realms") — server- authoritative roleplay (jobs, double-entry economies, property, police/EMS), proximity voice, a portable character identity (V6's Ori), composable executable mods (the productized Maya Forge: PubGrub resolution, semantic conflict merge, six-tier WASM sandbox, lock-file reproducibility), server meshing to 1000+ inhabitants, and a creator economy with a ≥70% direct-sale floor, a Fortnite-style 40% engagement pool, and — uniquely — **dependency revenue chains that pay framework authors automatically from the resolved lock file**. Trust and safety for executable UGC (Sekhmet) is platform- central and never realm-delegated: sandbox escape gates, fractureiser-class supply-chain defenses, PhotoDNA/Thorn-class child safety, day-one age assurance, and 29 adversarial CI gates. **The convergence thesis is correct and well-timed.** FiveM's serious-RP economy runs on gray-market payments and operator-root trust; Roblox is mid-reckoning on safety and pays creators a fraction of V7's floor; UEFN pays well but offers no persistent server-authoritative worlds; mod.io distributes but doesn't host. Nobody offers the combined stack, and GTA VI's arrival will destabilize the FiveM world V7 wants to inherit. Three design choices are genuinely beyond industry SOTA: the **hostile-operator trust boundary** (identity/money/child-safety platform-central; realm logic as capability-sandboxed WASM), **dependency revenue chains** (the ESX/QBCore maintainer class earns nothing today; paying them by construction is the sharpest wedge in the corpus), and **exit rights in writing** (signed realm exports, 24-month survival clause, sunset escrow — precisely what the post-Hytale climate makes resonant). **Three problems dominate.** (1) **Cold start without an anchor**: FiveM had GTA V; Roblox had 15 years; UEFN had Fortnite's audience. V7 has a client, a substrate, and no first-party world or resident population — and its "no FiveM compatibility, no ripped assets" stance (correct legally and architecturally) raises the migration cost for exactly the communities it courts. (2) **The runtime is libraries, not a service**: every Rust plane is health-endpoint-only with no network stack; meshing, voice, and persistence are proven as deterministic evals — the platform's hardest claims (1000+ meshed inhabitants, live community hosting) are its least real. (3) **The economics are thin by design**: a ≥70% creator floor plus a 40% engagement pool plus ~8%-margin hosting must fund Roblox-grade safety operations (vendors, human moderation, age assurance) that are fixed costs; the business model rests on unmeasured aggregate commerce. Verdict: **the right platform, specified with the industry's best safety and governance thinking, currently missing its audience, its accept loop, and its anchor world.** Ship a first-party flagship realm, turn spectating into the acquisition funnel RP already proved on Twitch, wire the safety vendors before any executable UGC goes live, and sequence meshing behind dedicated-server realism. --- ## 2. What V7 is (product identity) - A realm = base world + ordered Ixchel mod layers + Nàná job/economy config + membership policy + Eunomia governance charter + hosting binding, pinned in a content-addressed lock file; lifecycle Draft → Private → Listed → Verified → Archived; hosting continuum Solo → Listen (~16) → Dedicated (~256) → Meshed (1000+), promoted without rebuild. - Players inhabit persistent characters (Ori-backed; a realm owner cannot silently erase one), with jobs, licenses, property with meshed interiors, criminal/medical records, double-entry banking, proximity voice with radio/phone, and cross-realm crews with treasuries (Hera). - Creators get a four-tier authoring funnel (UE Editor → in-realm builder → visual-logic/TypeScript → AI-assisted, all emitting Ixchel layers), Wasmtime sandboxing with fuel/epoch budgeting, hot reload, deterministic replay debugging, a `moremi dev` single-binary loop, Collections, Loom world genomes, and git-like Variants with attribution and revenue links. - Platform-central trust: identity (opaque per-realm handles), real money (Aje; in-realm currency firewalled), client attestation, child safety, cross-realm bans — never delegated to operators; five-tier governance with a safety floor no vote can cross. --- ## 3. What is genuinely strong 1. **Dependency revenue chains** (§1) — computed from the lock file, so the QBCore-equivalent maintainer earns from every realm built on their framework, with the original creator never dropping below 70%. This is the single best creator-economy idea in the corpus, unmatched anywhere, and it targets the constituency (framework authors) whose defection decides RP-platform wars. 2. **The trust boundary.** "A realm operator is assumed potentially hostile" is the sentence FiveM cannot say and Roblox never needed to; V7's hybrid (community-hosted logic, platform-held identity/money/ safety) is the first architecture that makes community sovereignty and child safety compatible. The fractureiser-informed supply-chain posture (step-up re-publish verification, content-addressed surgical recall) is state of the art. 3. **Exit rights and data portability** — signed `realm-export/1` bundles, survival clause, sunset escrow, an independently verifiable CLI. After Hytale's cancellation and a decade of platform rug-pulls, "your world is yours in writing" is a marketing weapon competitors structurally cannot copy (their business models forbid it). 4. **Adversarial-gate culture at its best.** 29 seeded hostile fixtures as merge gates, each required to fail against a tampered implementation; a launch verifier that *cannot* report ready while any signoff is fixture-based (and currently, honestly, reports not-launch-ready); fixture-toy metrics explicitly labeled as such. This is the most trustworthy status reporting in the V-series. 5. **The Nàná roleplay domain is real code** — double-entry ledger with sink/faucet auto-balancing, jobs/grades/duties, property ACLs, CAD/MDT civic institutions — test-driven Rust, not spec. The serious-RP loop's hardest data problems are already solved. 6. **Voice done right**: positional occlusion-aware proximity voice with role-gated radio nets, honest scale budgets (≤40 streams/listener, p95 ≤250ms at 1,000 inhabitants), full text/caption accessibility parity, and a platform safety tap a realm cannot disable — pma-voice's feature set with none of its trust model. --- ## 4. Product gaps ### 4.1 No anchor world (the cold-start gap) A creator republic with no citizens pays no one. Every successful UGC platform bootstrapped on a first-party audience source; V7's plan leans on migration incentives (framework grants, white-glove moves, hosting discounts, 1.5× engagement multipliers) — supply-side subsidies for a demand-side problem. Communities follow audiences, not tooling. Two moves: - **Operate a first-party flagship realm** ("Mawu City"): a platform-run, serious-RP urban world — the NoPixel role played by the platform itself — that is simultaneously the reference implementation, the onboarding funnel, the live demonstration of meshing/voice/economy, and the content marketing engine. Its Nàná configuration and charter ship as the default template every new realm forks (Variants exists for exactly this). - **Decide the V5 relationship now.** V7 "extends V5's online backbone," but V5 is unshipped; if V5's urban world ever ships, it is the obvious anchor continent for realms. Either sequence V7 after a V5 slice, or build Mawu City as V7-native content and let V5 arrive later. What the corpus currently implies — a platform launching against the bare Maya substrate with no destination world — is the Hytale trap. ### 4.2 Spectating is the funnel, and it's underbuilt Serious RP is already a spectator business — NoPixel's audience is Twitch viewers, of whom a tiny fraction ever play. V7 has the pieces (delayed read-only spectating over the interest-managed delta stream, streamer mode, server-side replays/clips, Tier-2 web fallback for browse/spectate) but treats them as utilities. Assemble them into the acquisition loop: a streamer's overlay link → instant browser spectate of their realm (no install) → join the application queue in one click → whitelist onboarding. Add audience-participation primitives (viewer polls surfaced to in-realm events via the governance rails, clip-of-the-day per realm) and V7 owns the only funnel that converts RP's existing audience at zero acquisition cost. The realm browser listing "player count" should also list *viewer* count. ### 4.3 The runtime gap decides credibility sequencing Everything hard is a library with evals; nothing serves a socket. The right product consequence: **sequence public claims by hosting tier.** Ship and prove Solo → Listen → Dedicated-256 (which covers 95% of the FiveM-shaped market — few RP servers exceed 200 slots) before saying "meshing" in public. Star Citizen spent a decade making meshing real; V7's 2M-handoff eval is excellent groundwork and not a fleet. A dedicated-only launch with flawless voice/persistence/moderation beats a meshed promise that slips. Reserve Danu for the flagship realm's showcase district when it's real. ### 4.4 Safety cost vs. take-rate arithmetic V7 promises Roblox-grade safety (24/7 human moderation via Kuanyin, vendor fees for PhotoDNA/CSAI-Match/Thorn/age assurance, Lantern membership, bounties to $50k) on a platform take that is deliberately thin (≤30% of direct sales minus dependency shares; 60% of net after the engagement pool; hosting at ~8% margin). Roblox's effective ~65–70% take *funds* its trust apparatus. The docs model operator economics but not **safety unit economics**. Before GA: publish an internal cost-per-MAU safety model, and if the math fails, adjust *transparently* (e.g., a platform service fee on engagement-pool distributions, or safety costs deducted from eligible net before the 40% split — both defensible if stated up front; a silent floor erosion later would burn the republic's founding promise). ### 4.5 Governance over-supply, admin under-supply Eunomia ships six voting models and five sovereign tiers; real realm owners on day one need ban appeals, staff roles, whitelist queues, and an audit log — which exist — plus escalation muscle memory. Liquid democracy for a 40-person RP server is a feature nobody asked for yet. Ship **charter templates** (three archetypes: benevolent-dictator, staff-council, member-vote) with the voting machinery hidden until a realm opts into it; market governance depth to the mega-communities (10k+ member crews) where it actually bites. Meanwhile the truly scarce tool — **cross-realm moderation intelligence for operators** (is this applicant a serial griefer?) — exists as cross-realm reputation but needs a privacy-respecting operator surface with Themis-appealable provenance. ### 4.6 Industry SOTA checklist | Capability (2026 bar) | Bar-setter | V7 | |---|---|---| | Executable-UGC sandboxing | Roblox Luau VM | ✅✅ beyond (capability-typed WASM, fuel, zero-escape gate) | | Dependency/creator payouts | UEFN 40% pool | ✅✅ beyond (pool + 70% floor + dependency chains) | | Hostile-operator trust model | (nobody attempts) | ✅✅ novel; BYOC honestly weaker | | Server meshing in production | Star Citizen 4.0 (static) | ⚠️ eval-proven libraries; no fleet (§4.3) | | Serious-RP framework out of box | ESX/QBCore (community) | ✅ Nàná in tested Rust; needs content/templates | | Proximity voice at scale | pma-voice (FiveM) | ✅ spec + budgets; SFU unwired | | Console executable-UGC pipeline | ArmA Reforger + mod.io | ✅ modeled (cert-gated); multi-year bizdev reality | | Child safety / age assurance | Roblox (post-reckoning) | ✅✅ designed to the new bar; **0% vendor-wired** | | Exit rights / portability | (nobody) | ✅✅ category-defining | | Anchor first-party world | GTA V / Fortnite / Roblox corpus | ❌ **absent** (§4.1) | | Spectator-to-player funnel | Twitch RP culture (unowned) | ⚠️ pieces exist, unassembled (§4.2) | | Live network runtime | table stakes | ❌ health-endpoints only | | Federation/character corridors | (nobody) | ❌ headline pillar, zero code — descope or build | | IP Content-ID for 3D/audio | YouTube-class (2D only) | ⚠️ strong design paper; unbuilt | --- ## 5. Ideas that would make the product better 1. **Mawu City flagship realm** (§4.1) — first-party, charter-templated, the reference NoPixel; every subsystem's showcase and every new realm's fork parent. 2. **The spectator funnel** (§4.2) — browser spectate → application queue → whitelist onboarding; viewer counts in the realm browser; clip-of-the- day; audience polls into realm events. 3. **V6 agents as realm inhabitants.** The Aye Bridge already carries Ori beings into destination realms; let realm operators *hire* Egbe agents as persistent NPCs (shopkeepers, dispatchers, citizens) with capability-scoped incarnation. This solves the empty-realm cold start (a 12-person community's city feels alive), creates a V6↔V7 economy (agent stewards earn when their agents staff realms), and is a cross-product moat nobody — not Rockstar, not Roblox — can imitate. 4. **Framework-author residency program**: the dependency-revenue wedge (§3.1) aimed at named ESX/QBCore/Qbox/ox maintainers — guaranteed income floors for a year against porting their frameworks natively (grants already specced at $50k–150k; add revenue-floor guarantees and co-marketing). Win five maintainers and the migration story writes itself. 5. **Realm insurance as a product**: the world-state rollback machinery (PITR, surgical clawback, exploit runbooks) packaged as a visible operator promise — "exploit? your economy is recoverable to the minute" — with the Eunomia review for large rollbacks as the trust valve. FiveM operators live in fear of dupe-wipes; this is a switching reason. 6. **Creator analytics with retention lift** (`retentionLiftBasisPoints`) deserves headline billing: "know whether your mod makes players stay" — no modding ecosystem offers causal-ish content analytics; pair with an opt-in A/B slot in the compositor's variant machinery. 7. **A "clean-asset commons"**: the no-ripped-assets stance needs a carrot — commission/curate a CC-licensed base library (vehicles, furniture, clothing at RP-expected quality) seeded by the creator fund, dependency- revenue-eligible, so migrating communities rebuild their visual economy *inside* the payout system instead of mourning the gray one. 8. **Charter templates + governance progressive disclosure** (§4.5). 9. **Age-assurance as a public trust page**: V7 is built to the UK OSA/DSA bar before regulators ask; publish the architecture (vendor biometric deletion ≤60s, age-banded contact defaults, voice screening) as a parent-facing commitment — Roblox is spending 2026 apologizing; V7 can spend it promising. 10. **`moremi dev` in the browser**: the 30-minute zero-to-resource metric is the right DX north star; a web-hosted sandbox (Tier-2 + embedded SQLite build) that runs the tutorial realm without any install would put the first script moment inside the marketing site itself. --- ## 6. Criticisms and tweaks 1. **Do not ship executable UGC before the safety vendors are live.** The design is exemplary; the wiring is 0%. A single CSAM or fractureiser incident during a stand-in-scanner beta would be existential for a platform whose brand *is* the trust boundary. The launch verifier's no-fabrication invariant is the right guardrail — extend it to require vendor-integration proof (live PhotoDNA/Safer/EAC handshakes) as blocking evidence, not fixture passes. 2. **Descope federation from the launch narrative.** Character corridors and cross-realm economy bridges headline the identity story and have zero code. Cut them from V1 messaging; the Ori passport into V2–V6 is already a bigger story than realm-to-realm federation, and it at least has a tested bridge. 3. **Reconcile the glossary with disk** (nana Rust-only, Abundantia/ Eunomia framework-free TS, Pheme not a service) — small, but this corpus earns trust through precision; keep it. 4. **The Project ROME reliance** is strategically sensible and rhetorically risky: if Rockstar ships official RP with GTA VI, V7's window narrows to "the platform for what Rockstar won't allow" (original IP worlds, real creator payouts, exportable communities) — which is a *fine* position that the docs should articulate as the primary scenario, not the fallback. Plan for ROME being real. 5. **Take-Two adjacency hygiene**: the no-native-table, no-ripped-assets, clean-room posture is right; extend it to marketing language — never "FiveM successor," always "community worlds platform." The migration amnesty program should include IP-laundering checks (the clean-asset commons of §5.7 is the constructive answer). 6. **BYOC honesty**: the operator-data-protection paper concedes memory dumps and disk forensics sit outside WASM's coverage. Label BYOC realms in the browser (a trust badge distinction players can see), restrict them from age-banded minors entirely, and price the 15%/min-$19 tier to reflect the audit burden — the trust boundary should be legible at the join screen, not just in the architecture doc. 7. **Determinism concentration risk**: rollback netcode, mesh handoff, replay debugging, and PITR all stand on one deterministic core; the golden-replay gate guards it, but any engine upgrade (Wasmtime CVE → 48h patch posture → AOT recompile of the whole catalog) is a coordinated-risk event. Rehearse it: a quarterly "determinism fire drill" (forced Wasmtime bump + full-catalog recompile + golden-replay verification) should be an operational gate, not an incident-day improvisation. 8. **The 15 open `[ ]` items cluster exactly where players enter** (web client, builder, scripting compiler, hardened client, operator console) — the substrate is ahead of every surface. Rebalance: one full vertical slice (join a Listed dedicated realm from the UE client via Abundantia, talk on voice, hold a job, get paid, get moderated) is worth more than any additional eval gate. --- ## 7. Risks 1. **Cold start** (§4.1) — the platform graveyard (Hytale, Core, Crayta, Everywhere) is dug by exactly this; anchor world or bust. 2. **GTA VI / Project ROME** (§6.4) — both the opportunity's origin and its ceiling; position for coexistence. 3. **Safety incident pre-vendor-wiring** (§6.1) — existential. 4. **Safety economics** (§4.4) — the quiet solvency risk. 5. **Meshing overpromise** (§4.3) — sequence claims by tier. 6. **Migration refusal**: communities may simply stay on FiveM until Rockstar forces the issue; V7's window is event-driven — be ready to catch the diaspora (migration tooling rehearsed, amnesty live, §5.4 maintainers signed) *before* the triggering event, because platform choices happen in the two weeks after a shock. 7. **Governance abuse**: public steward/creator reputation + realm featuring creates capture incentives (review-brigading, curator corruption); the discovery-integrity paper knows this — its proposed eval gates (`discovery-integrity`) should be promoted from proposal to blocking before the browser ships. --- ## 8. Prioritized recommendations **P0 — sequence-defining** 1. First-party flagship realm + charter templates (§4.1, §5.1). 2. Vendor-wired safety as a blocking launch gate (§6.1). 3. Dedicated-tier vertical slice before meshing claims (§4.3, §6.8). 4. Framework-maintainer residencies signed (§5.4). 5. Safety unit-economics model published internally (§4.4). 6. Descope federation from launch messaging (§6.2). **P1 — the funnel wave** 7. Spectator funnel assembled end-to-end (§4.2). 8. V6-agents-as-inhabitants pilot (§5.3). 9. Clean-asset commons seeded (§5.7). 10. Browser-hosted creator tutorial (§5.10). 11. BYOC trust badging + minor restrictions (§6.6). **P2 — compounding** 12. Realm insurance productized (§5.5). 13. Retention-lift analytics headline + A/B slots (§5.6). 14. Age-assurance trust page (§5.9). 15. Determinism fire drills (§6.7). 16. Discovery-integrity gates promoted to blocking (§7.7). --- ## 9. Additional considerations (second pass) ### 9.1 Adult-content policy is a decision, not a default A large fraction of FiveM's serious-RP world includes 18+ themes; V7's corpus regulates illegal content thoroughly but is silent on *legal adult content* in realms. Silence resolves badly in both directions: a blanket ban forfeits communities, an unwritten allowance becomes a moderation crisis and a store-policy violation. Recommend an explicit adult-realm tier: age-verified access (the age-assurance rails already exist), excluded from discovery surfaces by default, ineligible for featuring/ engagement-pool boosts, clearly badged, and reconciled against Apple/Google/console store terms for the client apps. Decide it before the first whitelist application asks. ### 9.2 Gambling inside realms In-realm casinos are an RP staple, and V7's own docs flag the lootbox/ gambling counsel gap. The dangerous combination is realm-casino winnings adjacent to real-money rails (the currency firewall helps, but dependency revenue and engagement payouts create indirect cash-out gradients). Platform policy needed: in-realm gambling with in-realm currency only, no purchasable chips, regional gates where even simulated gambling is rated/restricted, and a hard structural rule that no gambling outcome can touch any Aje-settled value. This is a safety-floor candidate, not a charter option. ### 9.3 Discord is the adoption surface RP communities *operate* on Discord — applications, CAD/MDT links, staff escalation, ban appeals. A first-class Discord app (realm status embeds, application intake into the whitelist queue, appeal links into Themis) is cheaper than any migration grant and removes the most friction per dollar of anything in the go-to-market. Treat "works with your Discord" as a launch feature, not an integration afterthought. ### 9.4 Moderation verbs for RP culture The platform floor covers abuse; realm culture runs on norms the floor doesn't see — fail-RP, powergaming, metagaming. Give realm staff purpose-built tooling: RP-rule templates in Eunomia charters, incident annotations pinned to server-side replay clips (the evidence infrastructure exists), and escalation categories that distinguish norm-breaks from safety violations so platform queues aren't polluted with "he broke character." ### 9.5 Platform legal-entity readiness as a launch gate DMCA agent registration, DSA transparency-report obligations, OSA risk assessments, repeat-infringer policy publication, law-enforcement response procedures (realms will generate subpoenas) — the design papers cover the *architecture*; the corporate/legal operations checklist should be a blocking launch gate exactly like the 29 eval gates. A UGC platform is a legal entity with duties from day one of open registration. ### 9.6 Creator-payout tax operations at scale Engagement pools plus dependency chains produce thousands of micro-payees across jurisdictions — DAC7/1099-K thresholds, withholding, VAT-on- digital-services. Abundantia's compliance holds model the gates; the operational tax engine (and its vendor choice) needs to precede the first payout run, because retroactive tax cleanup at marketplace scale is the stuff of shutdowns. ### 9.7 Operator reputation, portable and visible Players choose realms partly on operator trustworthiness. A verified- operator profile (uptime history, moderation-SLA adherence, exports honored, BYOC vs platform-hosted badge) mirrors the steward-reputation concept and gives the trust boundary a *consumer-legible* surface at the join screen — extending §6.6's BYOC badging into a general operator trust system with Themis-appealable provenance. ### 9.8 Events as an economy layer Community tournaments and festivals inside realms (the V5 backbone already has bracket tooling) deserve platform-sanctioned event tooling: scheduled event listings in Abundantia, ticketed entry via the commerce rails, engagement-pool event multipliers routed through the anti-fraud graph. Big realms become venues; venues become a revenue line that doesn't touch the creator-floor economics. ### 9.9 Cross-portfolio notes Three siblings are V7's content supply chain: V6 agents as realm inhabitants (§5.3 — the flagship cold-start remedy), V5 city tiles as licensed base worlds (define the licensing boundary before the marketplace opens — mirrored in V5 §9.9), and V2/V4 modes as realm attractions via the same GameFeature composition. Sequence these seams deliberately: each one is a marketing beat ("your Egbe citizen now lives in Mawu") and a dependency to version-pin. ### 9.10 Publish the cost calculator Operator trust in the GSP market is won on predictable bills. The node-hour pricing exists; ship a public capacity/cost calculator (inhabitants × voice load × mesh nodes → monthly estimate) and commit to no-surprise billing (alerts at 80%, hard caps opt-in). The hosting business is deliberately thin-margin (§4.4) — its *strategic* return is trust, so instrument it for exactly that. ## 10. Monorepo integration & shared functionality (third pass) Audited against DOMAINS/, the 185-phase TODOS/ roadmap, `V_SERIES.md`, `V_SERIES_PLATFORM_CONSOLIDATION.md`, and package/crate evidence. ### 10.1 V7 is the GPS anchor — formalize the ownership `V_SERIES_PLATFORM_CONSOLIDATION.md` designates V7's hybrid trust model as the Game-Platform-Service tier's architectural anchor and makes V7 the M2 first native consumer. Given V7 is *itself* a platform product, this is more than consumption — **V7's team should own GPS delivery**: session auth on V1 identity, friends/parties/presence, matchmaking orchestration, leaderboards/seasons, telemetry, anti-cheat signal bus, replay storage, entitlements, and creator payouts (Abundantia generalized) are all services V7 must build anyway; building them multi-tenant by `gameId` from day one converts V7's platform work into the portfolio's consolidation and gives V7 the strongest possible internal mandate. Status reality: M1 (`libs/contracts/gps/`) does not exist — nine GPS domains without schemas — and V2's service sprawl grew from 50 to 91 while the plan sat unstarted. The exit criterion ("no new per-game impl of a GPS-owned domain") should be enforced by CI the way V7 enforces its 29 adversarial gates. ### 10.2 V7's lateral edges: one real, two fictional V7 owns the portfolio's only genuine cross-V dependency (Cargo path-deps on V6's `egbe-ori-service`/`ori-model` and Maya's `maya-world`) — proof the pattern works. But `V_SERIES.md` also claims V7 reuses "V3 protocol/audio/avatars" and "V5 online backbone & anti-cheat," and **neither edge exists in any package.json or Cargo.toml** — V7's `mawu-gateway` carries its own voice/SFU stack (the portfolio's third, after V3 and V6). Fix in the proven direction: consume the extracted V3 delivery/voice substrate (V3 §10.1) and the GPS anti-cheat bus (§10.1 above) via real workspace dependencies, and down-mark the DAG claims until then. A platform whose own dependency declarations are aspirational will struggle to police creators' dependency manifests. ### 10.3 Maya Forge: confirmed productization vehicle — mind the split `V_SERIES.md` settles the overlap this review's first pass inferred: root Phases 76/77/79 and V7 cover the same modding/governance ground, **V7 is the productization vehicle**, the `forge-*` crates were created as V7 work, and the abandoned OSHUN_VS_ROBLOX phase plan was folded into V7. Two follow-ups: (1) the production resolver/sandbox currently live inside `moremi-realm-server` while `libs/maya/forge-*` are early skeletons — finish the extraction so Maya-the-domain owns the engine pieces and V7 owns the product, or the next Maya consumer (V3/V5) forks again; (2) **Phase 78 (Forge AI modding companion)** and **Phase 134 (Forge 2D creative suite)** are roadmap items squarely inside V7's creator experience — claim or defer them explicitly. ### 10.4 Aphrodite twice over: consume the safety, adjudicate the runtime Two distinct Aphrodite findings. First, the **adult-content machinery**: Phase 20 shipped age-verified streaming, consent engineering, 2257 compliance, and performer-protection under Kuanyin — exactly the capability §9.1's adult-realm tier needs; consume it rather than rebuilding consent/verification inside Sekhmet. Second, the **orphaned game runtime**: `libs/aphrodite/` hosts 27 generic `game-*` packages and five `*-game-bridge` adapters (hathor/isis/lilith/nike/nyx) with **zero external consumers** — the repo's largest dead reuse asset. V7 is the only product whose Tier-2 web experiences and creator tooling could plausibly adopt parts of it (game-ecs, game-networking as the fallback runtime's spine); either that adoption is evaluated deliberately, or the portfolio should retire the island. What must not persist is the current state: the shared "inject any domain into a game" capability existing, unowned, under the adult domain's namespace where no SFW product will look for it. ### 10.5 Built-and-unwired trust assets Two completed roadmap items duplicate V7 plans on paper: **Phase 75 (Themis Universal Originality Shield)** is essentially the Content-ID-class system V7's IP-matching design paper proposes — before staffing that build, audit what Phase 75's artifacts actually deliver and wire the delta; and **Kuanyin (Phase 32)** provides the restorative- justice moderation layer V7's Kuanyin escalation path names — confirm the escalation path binds to the real service, not a placeholder. Same inheritance logic as V1's flywheel gate: **Phases 85–96** will want realm telemetry as training data; realm content is creator IP and player behavior — the training-consent gate must extend to operators and creators (a realm-charter clause), not just end users. ### 10.6 Cloud-gaming convergence **Phase 140 (sovereign cloud-gaming platform)** overlaps V7's realm streaming ambitions and V3's Pixel Streaming fleet (V3 §10.4). Three products, one GPU-streaming problem: whichever fleet ships first is the de-facto platform layer — the roadmap should name it (recommendation: V3's, as most complete) and V7 should consume it for spectate/Tier-2 rather than running a fourth worker fleet. ## 11. Closing note V7 has written down, with more rigor than any incumbent, how a community-worlds platform *should* work: operators assumed hostile and communities sovereign anyway; framework authors paid by construction; children protected by architecture rather than moderation heroics; and communities holding exit rights no platform has ever granted. What it has not yet built is the reason anyone shows up: a world, a crowd, and a socket that accepts their connection. The republic's constitution is ratified; it awaits territory and citizens. Found the capital city, open the spectator galleries, pay the framework elders their first residencies — and only then raise the flag that says "meshed."