Eve library

Eve — Elikem Granular Review: EGR.34 chunk-plan preview (planner egr-planner-6)

Evidence

Revision 19

On this page

Prepared 2026-10-04 by claude-workspace-ux-31-220-93-15 for eve-workspace-ux:EGR.34. Revision 3 of this record showed the plans of egr-planner-1 for ten real documents; Elikem reviewed them (first pilot, 2026-10-04) and asked that each finding be adopted in the next planner version. Revision 5 recorded his findings with their dispositions and showed the ten plans under egr-planner-2. Revision 9 answered review 15e0b775 of revision 6 with egr-planner-3, and revision 11 answered review 6d1313f5 of revision 9. Revision 17 adds egr-planner-6, which answers review 40c19fe4 (below). Revision 16 answered Elikem delegate decision f3a71d16 on revision 12: every “Opens with” excerpt is literal source text in inline code, so a link copied from a source is never a live link here. It shows the ten plans regenerated by egr-planner-4 (commit 96bf89f664; egr-claims-1, egr-estimator-1 unchanged) for his second pilot. Nothing was stored: no campaign, plan or decision exists. Estimates are planning aids, not quotas.

For Elikem (second pilot): do the five changes below read right on these documents? Reply with findings, or "adopt as is".

Findings from the first pilot and their dispositions#

# Elikem's finding Disposition What egr-planner-4 does
1 Skip boilerplate chunks: metadata headers, revision histories, reference lists (ADR-0004/0005 chunks 1 and 7; content workbenches chunk 26) Adopted A metadata header (a run of Status:, Date:, Authors: … pairs at the top), and sections titled only with furniture (status, revision or version history, changelog, references, sources, research documents, bibliography, see also; "Sources & References", "Research and sources (2026-09-27)") are excluded with their reason. They are disclosed ("Not asked" below; synthesis and progress count them as exclusions) and never asked about.
2 Machine-check generated tables: ask whether the rule is right, not the rows (architecture overview's Domain Responsibilities Matrix) Adopted A table is generated only when a passage before it, in the same section and not itself a table, says so ("Generated from …", "auto-generated", "do not edit") and names its generator or check in code (…). That passage and the table header become the question "Is the rule that generates this table right?"; the rows are excluded as "generated rows, checked by" that check. Wording inside an ordinary table never makes it generated. Generated tables found: Oshun Monorepo — Architecture Overview, “Domain Responsibilities Matrix” (2.8 min; generated rows, checked by node tools/domains/check-registry.mjs).
3 The question follows the section's role (content workbenches chunks 4 and 26) Adopted Method notes and reference material (how this was researched, method, background, glossary, appendices, related decisions, where to look next) are asked "Anything wrong or missing here?". Other sections keep the question of their document's kind.
4 Don't join unrelated sections ("Nyx · Veritas"; "Event Flow · API Gateway & BFF layer · Build system") Adopted Short neighbours join only when their titles share a subject word (generic words such as matrix, table, system or overview do not count): "Retention" joins "Retention exceptions"; Nyx and Veritas, and Event Flow, API Gateway and Build system, are separate chunks. A heading never stands alone as a chunk: a run of headings joins the section it introduces, and headings at the end join the section before them. Heading-only chunks in these plans: 0.
5 Hide claim counts when unknown dominates (ADR-0004 chunk 2: 1/5/107) Adopted Each chunk carries its counts (headings, which assert nothing, are not counted) and shows them only when unknown is not the majority; otherwise "hidden" below, and the preview omits them. The labels themselves stay one click away.
— Estimates about right; keep the estimator Adopted (no change) egr-estimator-1 is unchanged. Totals move two ways: excluded material is no longer read as a question (less time), and unrelated sections now stand apart, each chunk carrying its own decision minute (more).
— Stale status text is worth surfacing as a claim (content workbenches chunk 1: "draft for Elikem's approval" on an approved document) Adopted When Eve has a document approved but its status text says draft, for approval, in review or proposed, the chunk carries a stale-status flag with the quoted text (shown below and in the preview). Found: 2 (Lilith content workbenches — state of th…, chunk 1; Lilith V1.3 — the personal agent: scope …, chunk 1).

Reviews 15e0b775 and 6d1313f5, and what changed#

Finding egr-planner-4
Lilith content workbenches, chunk 27: the CR-1 to CR-9 decision table was excluded as generated because CR-3 says its artwork is "generated from one style kit", with the check named "one" Generation is read only from a passage before the table that names its generator or check in code; the CR table is primary again and asked as decisions. A regression test covers generation wording inside an ordinary decision table.
OSHUN Unified Consumer PRD, chunks 19, 38 and 45: heading-only chunks Heading-only groups join the section they introduce (or, at the end, the one before). A regression test covers runs of same-depth headings.
Review 6d1313f5, OSHUN Unified Consumer PRD, chunks 26–27 and 44–45: the introductions of sections 15 and 22 attached to the sections before them, through the neighbouring-title join egr-planner-4: an introduction never joins the section before it, whatever words their titles share; “15. Image, Audio, Video, Avatar, and Model System Requirements” now opens the chunk of 15.1, and “22. Non-Functional Requirements” the chunk of 22.1 (as does “21.” with 21.1). A regression test covers numbered sections at one depth.

Across the ten documents: 247 chunks, 195 passages not asked (boilerplate, metadata, generated rows), 1 generated-table questions, 5 method or reference questions, claim counts hidden on 130 chunks.

All ten plans together: about 929 minutes (revision 12: about 929; egr-planner-3: about 927; egr-planner-1: about 921).

Review 40c19fe4 and egr-planner-6#

Finding egr-planner-6 (commit b5608c36ec)
F1: generation wording could exclude an ordinary decision table (the CR-1 to CR-9 rulings, through CR-3's "generated from one style kit") Since egr-planner-4 the CR table is primary. egr-planner-6 makes the rule an explicit declaration: rows are excluded as generated only when the passage directly before the table names the source it is generated from and the check that verifies it, each in code. Prose naming no check, wording inside the table, or a declaration a paragraph away leaves the rows under review. Regression tests cover each case.
F2: a revision with only boilerplate became an empty "Opening" decision; a title over an excluded Status section a heading-only one A group with no substantive primary unit joins the substantive chunk after it (or before it at the end); a revision with nothing to review plans no chunk and discloses every unit, each with its reason. Generated plans are validated before they are proposed or stored.

Measured on the ten pinned revisions listed below (2026-10-08, egr-planner-6 against egr-planner-5): unchanged. 247 chunks, about 929 minutes, every plan valid, 0 heading-only chunks, 0 revisions with nothing to review; the CR-1 to CR-9 table is primary in "Lilith, characters and the creative studio"; the one generated table found is still the Architecture Overview's Domain Responsibilities Matrix (48 generated rows, checked by node tools/domains/check-registry.mjs), whose declaration names its check. The listings below are therefore the plans of egr-planner-6 as well (chunk identities aside, which carry the planner version).

Lilith content workbenches — state of the art for V1.0 to V1.4: scope (2026-09-29)#

Revision 18 · prd reviewed as prd · approved · 28 chunks · about 138 minutes in all (revision 12: 28 chunks, about 138 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 Lilith content workbenches — state of the art for V1.0 to V1.4: scope (2026-09-29) Is this still what you want built? 3 min (5) 10 / 1 / 0 1: metadata header Stale status: “Status: **draft for Elikem's approval; the §9 decisions were settled by Elikem on 2026-09-”. An independent read-only review (Codex, 2026-09-29) was folded in before filing. It checked the claims about c
2 0. The brief (Elikem, 2026-09-28, verbatim) Is this still what you want built? 2.8 min (5) 2 / 0 / 0 — > so this time focusing on the workbenches for creating content for all of oshun V1.0 to V1.4 please do thorou
3 1. Summary Is this still what you want built? 4.7 min (5) 13 / 2 / 0 — **The finding in one line:** Lilith's rooms can show content, but only Tara can author it. Even Tara authors i
4 2. What exists today (code and board, 2026-09-28) · 2.1 Found during this research: the li Is this still what you want built? 8.1 min (10) 4 / 3 / 1 — Measured from the repository at 'main' and the Eve board on 2026-09-28. Detail is in the research documents; t
5 3. How this was researched Anything wrong or missing here? 4.3 min (5) 5 / 1 / 0 — - **Inventory first.** Code search over the rooms' content models, studios, generation paths and persona layer
6 4. Principles (binding on every task in this family) Is this still what you want built? 4.6 min (5) 13 / 2 / 0 — 1. **Structure before prose.** Every content item is composed from typed domain models: a practice, a sky clai
7 5. What a state-of-the-art content workbench is Is this still what you want built? 4.6 min (5) 1 / 1 / 0 — Across all six rooms and the creative studio, the research found the same twelve stages. Leading producers do
8 6. Scope Is this still what you want built? 3.7 min (5) 1 / 1 / 0 — 315 tasks and 8 epics, in eight pillars. Task keys below are the board keys in family 'content-workbenches'. E
9 6.1 Shared studio core [CO] Is this still what you want built? 4.9 min (5) 2 / 1 / 0 — Six rooms kept rediscovering the same needs, so they are built once. The research found that state-of-the-art
10 6.2 Tara: contemplative content [TA] Is this still what you want built? 8.1 min (10) 11 / 1 / 0 — Research: Eve document 13d4487f-57de-4523-81d6-e919e7b2afba.
11 6.3 Nyx: sky content [NX] Is this still what you want built? 8.5 min (10) 8 / 1 / 2 — Research: Eve document 788820e4-8a29-450b-b31c-86104a08c7a4.
12 6.4 Arete: coaching content [AR] Is this still what you want built? 7.4 min (10) 11 / 0 / 0 — Research: Eve document c479a954-f5e0-43ff-91f6-02520dbf7348.
13 6.5 Nisaba: text and study content [NI] Is this still what you want built? 6.9 min (10) 9 / 2 / 1 — Research: Eve document 8563ce32-96b4-45d3-af99-5115633fb753.
14 6.6 Metis: learning content [ME] Is this still what you want built? 8.5 min (10) 13 / 0 / 0 — Research: Eve document b325c39f-6ba1-4b49-98ba-c2d062dfde80.
15 6.7 Veritas: news and analysis content [VE] Is this still what you want built? 8.2 min (10) 13 / 0 / 0 — Research: Eve document c15926ca-fd59-414b-a3bf-614c4551a235.
16 6.8 Lilith, characters and the creative studio [CR] Is this still what you want built? 8.6 min (10) 10 / 1 / 1 — Research: Eve document 17508fb3-47c1-453a-a3f7-8bb1cf1fe950.
17 7. Sequencing and milestones Is this still what you want built? 4.8 min (5) 1 / 0 / 1 — / Milestone / Scope / Release it serves / /---/---/---/ / **M0 Foundations and urgent fixes** / The architectu
18 8. Relationship to other plans Is this still what you want built? 2.8 min (5) 3 / 0 / 1 — - **'v1-domain-workbenches'** - The Metis (Phase M) and Veritas (Phase V) ledgers stay as written. ME01 and VE
19 9. Decisions only Elikem can make Is this still what you want built? 1.5 min (5) 2 / 0 / 0 — The research passes raised 70+ questions. They reduce to the decisions below, each with a recommendation.
20 Across all rooms Is this still what you want built? 4.3 min (5) 1 / 0 / 0 — / # / Decision / Recommendation / /---/---/---/ / X-1 / Fund paid external reviewers per room: meditation teac
21 Tara Is this still what you want built? 3.4 min (5) 1 / 0 / 0 — / # / Decision / Recommendation / /---/---/---/ / TA-1 / Intense breathwork (hyperventilation, retention) / Ex
22 Arete Is this still what you want built? 3.1 min (5) 0 / 1 / 0 — / # / Decision / Recommendation / /---/---/---/ / AR-1 / 'libs/arete/seven-habits' (FranklinCovey trademark, C
23 Nisaba Is this still what you want built? 3 min (5) 1 / 0 / 0 — / # / Decision / Recommendation / /---/---/---/ / NI-1 / KJV for UK members / Serve a public-domain translatio
24 Nyx Is this still what you want built? 2.8 min (5) 1 / 0 / 0 — / # / Decision / Recommendation / /---/---/---/ / NX-1 / Astrology / Cultural history only: no horoscopes, pre
25 Veritas Is this still what you want built? 3.4 min (5) 1 / 0 / 0 — / # / Decision / Recommendation / /---/---/---/ / VE-1 / Outlet ratings: build or license / Build our own publ
26 Metis Is this still what you want built? 3.1 min (5) 1 / 0 / 0 — / # / Decision / Recommendation / /---/---/---/ / ME-1 / Under-18 learners and sales to schools in England / A
27 Lilith, characters and the creative studio Is this still what you want built? 4 min (5) 1 / 0 / 0 — / # / Decision / Recommendation / /---/---/---/ / CR-1 / Who owns Lilith's character? / One named human editor
28 10. Risks Is this still what you want built? 4.7 min (5) 1 / 0 / 0 3: boilerplate: Research documents / Risk / Where it bites / Mitigation in this family / /---/---/---/ / **Fabrication at scale.** Cheap models r

Lilith V1.3 — the personal agent: scope (2026-09-27)#

Revision 47 · prd reviewed as prd · approved · 38 chunks · about 182 minutes in all (revision 12: 38 chunks, about 182 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 1. Summary Is this still what you want built? 2.1 min (5) 2 / 0 / 0 1: metadata header Stale status: “Status: draft for Elikem's approval.”. V1.3 turns Lilith from a host that acts inside her own rooms into a personal agent that acts across the member
2 2. Where Lilith stands today (V1.0 code, 2026-09-27) Is this still what you want built? 6 min (10) 0 / 10 / 0 — / Area / Today / Where / /---/---/---/ / Assistant / Lilith, "host of the house"; real model through OpenRoute
3 3. The field in September 2026 Is this still what you want built? 9.4 min (10) 7 / 0 / 1 — / Product / What it does for an individual / Surfaces / Price / Lesson for Lilith / /---/---/---/---/---/
4 3. The field in September 2026 Is this still what you want built? 8 min (10) 15 / 2 / 0 — / **xAI Grok Bot** (beta 2026-08-11) / Always-on agents on a cloud computer with a browser and shell: they sig
5 4. Positioning and differentiators Is this still what you want built? 4 min (5) 14 / 0 / 1 — **Lilith V1.3 is the personal agent that does everything Muse does, guards your wellbeing while doing it, and
6 5. Principles (binding on every V1.3 task) Is this still what you want built? 3.7 min (5) 10 / 1 / 0 — 1. **Nothing fabricated.** The repository's quality bar holds: every capability either works against the real
7 6. Scope Is this still what you want built? 1.2 min (5) 1 / 0 / 0 — Each pillar below is a board epic in family 'lilith-v1-3' (key in brackets). The atomic tasks under each carry
8 6.1 Programme and release gate [G] Is this still what you want built? 2.3 min (5) 2 / 1 / 0 — - A fail-closed V1.3 capability gate in 'release-scope.ts' (same pattern as payment rails: unknown ids refuse)
9 6.2 Agent workspace and containment [W] Is this still what you want built? 3.5 min (5) 9 / 0 / 0 — - **Per-member secure workspace**: an isolated sandbox (Firecracker/gVisor-class microVM or container with sec
10 6.3 Connectors and permissions [C] Is this still what you want built? 2 min (5) 4 / 0 / 0 — - Connector framework: OAuth 2.1 + PKCE, incremental scopes, refresh, revocation webhooks, health checks, per-
11 6.4 Actions in the world [A] Is this still what you want built? 6.1 min (10) 13 / 0 / 0 — - **Email**: triage, summaries, drafts in the member's voice, send-with-approval, follow-up tracking; an agent
12 6.5 Commerce and payments [P] Is this still what you want built? 3.4 min (5) 5 / 2 / 0 — - Stripe Link for agents: OAuth wallet access, per-task one-time cards and Shared Payment Tokens scoped by amo
13 6.6 Proactivity and background work [R] Is this still what you want built? 5.8 min (10) 9 / 0 / 1 — - Background tasks that run with the member's devices off, a concurrency limit per plan, pause/resume/cancel,
14 6.7 Memory and personalization [M] Is this still what you want built? 1.7 min (5) 6 / 1 / 0 — - Flow labels on every memory (contextual integrity) enforced by the policy engine; health and mood memories s
15 6.8 Wellbeing intelligence [H] Is this still what you want built? 7 min (10) 10 / 0 / 0 — - Health data in: HealthKit and Health Connect (native bridges), Oura, WHOOP, Garmin, Withings, Fitbit; consen
16 6.9 Surfaces and devices [S] Is this still what you want built? 2.9 min (5) 6 / 2 / 0 — - Web/PWA and native iOS/Android (V1.1 base): Live Activities / ongoing notifications for running tasks, widge
17 6.10 Voice, realtime and presence [V] Is this still what you want built? 4.1 min (5) 7 / 1 / 0 — - Full-duplex speech-to-speech with barge-in, proactive silence and hand-off to a reasoning model mid-conversa
18 6.11 Bring your own model [B] Is this still what you want built? 4.3 min (5) 4 / 3 / 0 — - **Connect any model** from one "Models" page: - OpenRouter "connect your account" (OAuth PKCE to '/auth', ex
19 6.12 Model lab — routing, ensembles and cost [L] Is this still what you want built? 4 min (5) 5 / 1 / 0 — - **Model catalogue** synced from OpenRouter's model API and provider price lists (the method in Eve document
20 6.13 Model lab — evals and benchmarking [E] Is this still what you want built? 3.2 min (5) 6 / 1 / 0 — - **Personal eval suite** mined (with consent) from the member's own history: recurring task types become case
21 6.14 Model lab — skills, optimisation and fine-tuning [K] Is this still what you want built? 2.6 min (5) 6 / 1 / 0 — - **Agent Skills standard** (SKILL.md folders; 'name' ≤ 64, 'description' ≤ 1024, optional 'allowed-tools'; pr
22 6.15 Glass box — reasoning, traces and inspection [X] Is this still what you want built? 3.2 min (5) 5 / 2 / 0 — - **Trajectory log**: append-only, per session, with inspect, fork, replay, resume and export (OpenTelemetry G
23 6.16 Household, family and friends [F] Is this still what you want built? 2 min (5) 5 / 0 / 0 — - A household agent with members, roles and per-member sharing rules (per sender, per calendar, per data class
24 6.17 Developer ecosystem [D] Is this still what you want built? 1.5 min (5) 3 / 1 / 0 — - A connector platform: remote MCP servers with OAuth, review (function, security, legal), signing, versioning
25 6.18 Trust, safety and compliance [T] Is this still what you want built? 5.7 min (10) 8 / 0 / 1 — - Prompt-injection defence in depth: a privileged planner and a quarantined parser for untrusted content (the
26 6.19 Packaging and pricing [Q] Is this still what you want built? 2.5 min (5) 5 / 1 / 0 — - Free tier with a weekly compute budget; paid tiers by compute, not by task count; a BYOK / self-hosted / on-
27 6.20 Installable apps and distribution [I] Is this still what you want built? 5.6 min (10) 4 / 0 / 1 — Where the field stands (September 2026): Claude's desktop app reached Linux in beta on 2026-06-30 (Ubuntu/Debi
28 7. Release plan Is this still what you want built? 7.4 min (10) 2 / 1 / 0 — V1.3 ships in four milestones. Each milestone ends with the red-team pass ('T'), the personal-agent benchmark
29 8. Non-goals Is this still what you want built? 1.7 min (5) 6 / 1 / 0 — - Oshun hardware (a pendant or glasses of our own). Lilith runs on the devices members already own.
30 9. Success measures Is this still what you want built? 1.9 min (5) 8 / 0 / 0 — - **Task success** on Lilith's personal-agent benchmark and on each member's own suite; published per release.
31 10. Decisions for Elikem Is this still what you want built? 6.2 min (10) 12 / 1 / 0 — 1. **Launch regions** — US-first (as Muse and Spark did) or UK/EU at launch with GDPR and Art. 50 from day one
32 11. Risks Is this still what you want built? 5.5 min (10) 1 / 0 / 0 35: boilerplate: Research and sources (2026-09-27) / Risk / Mitigation / /---/---/ / Prompt injection leads to a harmful action / Planner/quarantine split, Senti
33 13. Board map (family lilith-v1-3) Is this still what you want built? 8 min (10) 8 / 0 / 0 — Every task starts as 'later'; the family stays 'unstarted' until this scope is approved. Each task body carrie
34 13. Board map (family lilith-v1-3) Is this still what you want built? 8.6 min (10) 6 / 0 / 0 — / Task / M / Effort / Depends on / /---/---/---/---/ / 'A01' Email: triage, drafts and send with approval / M1
35 13. Board map (family lilith-v1-3) Is this still what you want built? 7.7 min (10) 6 / 0 / 0 — / Task / M / Effort / Depends on / /---/---/---/---/ / 'M01' Flow labels on every memory / M1 / M / W03 / / 'M
36 13. Board map (family lilith-v1-3) Is this still what you want built? 10 min (10) 7 / 0 / 0 — / Task / M / Effort / Depends on / /---/---/---/---/ / 'I01' Platform matrix and support policy / M0 / S / — /
37 13. Board map (family lilith-v1-3) Is this still what you want built? 9.5 min (10) 9 / 0 / 0 — **E — Model lab — evals and benchmarking** ('lilith-v1-3:E')
38 13. Board map (family lilith-v1-3) Is this still what you want built? 7.6 min (10) 6 / 0 / 0 — / Task / M / Effort / Depends on / /---/---/---/---/ / 'D01' Connector platform with review / M3 / L / C09, K0

Yemaya Platform#

Revision 1 · prd reviewed as prd · draft · 46 chunks · about 257 minutes in all (revision 12: 46 chunks, about 257 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 Yemaya Platform Is this still what you want built? 1.8 min (5) hidden (1 of 1 unknown) — > **Legacy Documentation:** These docs are from the original standalone Yemaya > project before it was absorbe
2 Product Requirements Document (PRD) · The Premier Industry-Leading Platform for End-to-End Is this still what you want built? 1.2 min (5) 1 / 0 / 0 — **Version:** 2.0 **Date:** December 2025 **Status:** Draft - SOTA Update
3 Table of Contents Is this still what you want built? 7.8 min (10) hidden (15 of 15 unknown) — 1. [Executive Summary](#1-executive-summary)
4 1. Executive Summary · Key Differentiators · Target Users Is this still what you want built? 4.2 min (5) hidden (12 of 12 unknown) — Yemaya is envisioned as the world's most comprehensive, state-of-the-art platform for creating movies, games,
5 2. Vision & Mission · Vision Statement · Mission · Core Principles Is this still what you want built? 2.8 min (5) hidden (7 of 8 unknown) — _"To democratize AAA-quality creative production by unifying the world's best open-source tools with cutting-e
6 3. Market Analysis · Industry Context (2025) · Competitive Landscape Is this still what you want built? 6.6 min (10) hidden (23 of 23 unknown) — **Game Development:**
7 4. Platform Architecture · High-Level Architecture · Technology Stack Is this still what you want built? 4.3 min (5) hidden (2 of 2 unknown) — ''' ┌─────────────────────────────────────────────────────────────────────────┐ │ YEMAYA PLATFORM ARCHITECTURE
8 5. Core Modules · 5.1 Pre-Production Suite · 5.1.1 Script & Story Development · 5.1.2 Conc Is this still what you want built? 10 min (10) hidden (12 of 12 unknown) — The Pre-Production Suite handles everything before cameras roll or game engines compile.
9 5.2 3D Creation & Game Engine Core Is this still what you want built? 1.1 min (5) hidden (1 of 1 unknown) — The heart of Yemaya: professional 3D creation and interactive runtime powered by best-in-class open source eng
10 5.2.1 Engine Integration Layer Is this still what you want built? 9.8 min (10) hidden (9 of 10 unknown) — **Primary Engines:**
11 5.2.2 3D Asset Creation Pipeline Is this still what you want built? 5.2 min (10) hidden (4 of 4 unknown) — / Stage / Tool / AI Capabilities / / ---------------------- / ------------------------------------------------
12 5.2.3 Character Creation · 5.2.4 Environment & World Creation Is this still what you want built? 6.7 min (10) hidden (22 of 22 unknown) — **MetaHuman Integration:**
13 5.2.5 Game-Specific Features Is this still what you want built? 2.9 min (5) hidden (2 of 2 unknown) — **Core Game Systems:**
14 5.3 Film Production Suite Is this still what you want built? 1.1 min (5) hidden (1 of 1 unknown) — Comprehensive film production tools from virtual production to final delivery.
15 5.3.1 Virtual Production Stage Is this still what you want built? 2.4 min (5) hidden (9 of 9 unknown) — **LED Wall Virtual Production:**
16 5.3.2 Motion Capture Integration Is this still what you want built? 5.6 min (10) hidden (28 of 29 unknown) — **Supported Systems:**
17 5.3.3 Video Editing & Post-Production Is this still what you want built? 5.2 min (10) hidden (13 of 13 unknown) — **DaVinci Resolve 20 Integration (August 2025):**
18 5.3.4 VFX & Compositing Is this still what you want built? 2.8 min (5) hidden (7 of 7 unknown) — **Compositing Pipeline:**
19 5.3.5 AI Video Generation Is this still what you want built? 5.2 min (10) hidden (10 of 10 unknown) — **Integrated AI Video Tools (2025 SOTA):**
20 5.4 Audio & Music Production Suite Is this still what you want built? 1.1 min (5) hidden (1 of 1 unknown) — Complete audio production from scoring to final mix.
21 5.4.1 Digital Audio Workstation Is this still what you want built? 2.4 min (5) hidden (7 of 7 unknown) — **Ardour (Core DAW):**
22 5.4.2 Music Composition Is this still what you want built? 6.8 min (10) hidden (17 of 17 unknown) — **Composition Tools:**
23 5.4.3 Sound Design Is this still what you want built? 2.3 min (5) hidden (7 of 7 unknown) — **Sound Design Pipeline:**
24 5.4.4 Voice Production Is this still what you want built? 7.8 min (10) hidden (14 of 14 unknown) — **Voice Pipeline:**
25 5.4.5 Adaptive Audio System Is this still what you want built? 2.7 min (5) hidden (2 of 2 unknown) — **Game Audio Features:**
26 5.5 AI Agent Framework Is this still what you want built? 1.2 min (5) hidden (1 of 1 unknown) — The AI Agent Framework is Yemaya's most innovative component: a comprehensive system for AI-assisted and fully
27 5.5.1 Agent Architecture · 5.5.2 Agent Types Is this still what you want built? 7.6 min (10) hidden (8 of 8 unknown) — **Multi-Agent System Design:**
28 5.5.3 Agent Framework Stack Is this still what you want built? 4.1 min (5) hidden (4 of 4 unknown) — **Core Technologies:**
29 5.5.4 Autonomous Creation Modes Is this still what you want built? 4.5 min (5) hidden (4 of 4 unknown) — **Three Operational Modes:**
30 5.5.5 NPC AI System Is this still what you want built? 5.8 min (10) hidden (31 of 31 unknown) — **LLM-Powered NPCs for Games (2025 SOTA):**
31 5.6 Asset Management & Pipeline · 5.6.1 Version Control · 5.6.2 Digital Asset Management ( Is this still what you want built? 7.2 min (10) hidden (13 of 13 unknown) — **Dual-System Approach:**
32 5.7 Cultural Research & Localization · 5.7.1 Cultural Research Tools · 5.7.2 Localization Is this still what you want built? 7.7 min (10) hidden (14 of 14 unknown) — **Research Database:**
33 5.8 XR & Spatial Computing Suite · 5.8.1 Apple Vision Pro / visionOS Development · 5.8.2 C Is this still what you want built? 7.5 min (10) hidden (31 of 31 unknown) — Yemaya provides comprehensive support for extended reality (XR) development, including virtual reality (VR), a
34 5.9 Real-Time Collaboration Hub · 5.9.1 Video Collaboration Platforms · 5.9.2 Production M Is this still what you want built? 8.7 min (10) hidden (16 of 16 unknown) — Yemaya provides comprehensive real-time collaboration tools for distributed creative teams, addressing the ind
35 6. AI Integration Architecture · 6.1 LLM Integration Layer Is this still what you want built? 6.3 min (10) hidden (10 of 11 unknown) — **Supported LLM Providers (2025 SOTA):**
36 6.2 Generative AI Integration Is this still what you want built? 8.8 min (10) hidden (12 of 12 unknown) — **Image Generation (2025 SOTA):**
37 6.3 AI Safety & Ethics Is this still what you want built? 2.8 min (5) hidden (8 of 8 unknown) — **Content Filtering:**
38 7. Human-AI Collaboration Framework · 7.1 Collaboration Modes · 7.2 Iterative Refinement · Is this still what you want built? 8.3 min (10) hidden (9 of 10 unknown) — **The AIdeation Model (CHI 2025 Inspired):**
39 8. Theory Integration Layer · 8.1 Game Design Theory · 8.2 Film Theory · 8.3 Interactive N Is this still what you want built? 11 min (15) hidden (12 of 12 unknown) — **MDA Framework Implementation:**
40 9. Emerging Technologies Integration · 9.1 3D Gaussian Splatting & Radiance Fields · 9.2 O Is this still what you want built? 11.9 min (15) hidden (52 of 52 unknown) — **Overview:** 3D Gaussian Splatting (3DGS) represents a breakthrough in real-time 3D capture and rendering, ac
41 10. Technical Requirements · 10.1 System Requirements · 10.2 Platform Support · 10.3 API & Is this still what you want built? 8.9 min (10) 9 / 0 / 0 — **Development Workstation:**
42 11. Open Source Foundation · 11.1 Core Open Source Components · 10.2 Yemaya Open Source Co Is this still what you want built? 6.1 min (10) hidden (4 of 5 unknown) — / Component / Tool / License / Role in Yemaya / / ------------------ / ------------- / ----------------------
43 12. Deployment Models · 12.1 Deployment Options · 12.2 Licensing Tiers · 12.3 AI Credit Sy Is this still what you want built? 8 min (10) hidden (5 of 5 unknown) — / Model / Description / Best For / / ------------------- / --------------------------------------- / ---------
44 13. Success Metrics · 13.1 Platform Metrics · 13.2 Quality Metrics · 13.3 Impact Metrics Is this still what you want built? 6.1 min (10) hidden (3 of 3 unknown) — / Metric / Target (Year 1) / Target (Year 3) / / --------------------- / --------------- / --------------- / /
45 14. Roadmap · Phase 1: Foundation (Months 1-6) · Phase 2: Production Tools (Months 7-12) · Is this still what you want built? 6.2 min (10) 4 / 0 / 0 — ''' CORE PLATFORM ├─ Yemaya Core architecture ├─ Blender integration (deep) ├─ Godot integration (deep) ├─ Unr
46 15. Appendices · Appendix A: Glossary · Appendix B: Competitive Analysis · Appendix C: Int Anything wrong or missing here? 8.1 min (10) hidden (18 of 18 unknown) 80: boilerplate: Document History; boilerplate: Sources & References / Term / Definition / / ----------- / ------------------------------------------------------- / / **AAA** / Hi

OSHUN Unified Consumer, Admin, and Core Systems PRD#

Revision 1 · prd reviewed as prd · draft · 58 chunks · about 107 minutes in all (revision 12: 58 chunks, about 107 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 OSHUN Unified Consumer, Admin, and Core Systems PRD Is this still what you want built? 1.5 min (5) hidden (7 of 7 unknown) 1: metadata header - 'docs/releases/v1/reports/unified-product-gap-analysis-and-nisaba-integration-report-2026-03-22.md'
2 Table of Contents Is this still what you want built? 1.9 min (5) hidden (27 of 27 unknown) — 1. Executive Summary
3 1. Executive Summary Is this still what you want built? 3 min (5) hidden (33 of 34 unknown) — OSHUN V1 is the fully unified contemplative intelligence ecosystem.
4 2. Product Vision · 3. Product Thesis Is this still what you want built? 3 min (5) hidden (18 of 24 unknown) — OSHUN should become the most compelling contemplative intelligence product in the market: a product that helps
5 4. Scope Is this still what you want built? 1.8 min (5) 18 / 0 / 0 — This PRD defines the end-state V1 requirements for:
6 5. Product Principles · 5.1 Philosophy First · 5.2 Tara as the Center · 5.3 Cross-Domain D Is this still what you want built? 2.4 min (5) 6 / 0 / 2 — OSHUN must feel like one philosophy expressed through multiple experiences, not several unrelated apps gathere
7 6. Users and Stakeholders · 6.1 Customer Personas · The Seeker · The Builder of Character Is this still what you want built? 2.7 min (5) hidden (25 of 25 unknown) — Wants meditation, rituals, meaning, and a calmer relationship with technology.
8 7. Product Surface Map · 7.1 Customer Surfaces · OSHUN Mobile · OSHUN Web · 7.2 Admin Surf Is this still what you want built? 2.1 min (5) hidden (27 of 27 unknown) — The primary habit and daily relationship surface.
9 8. Experience Narrative · 8.1 What the Product Should Feel Like · 8.2 Daily Flow Model · M Is this still what you want built? 4.2 min (5) hidden (78 of 86 unknown) — OSHUN should feel serene, capable, intelligent, and deeply integrated.
10 10. Cross-Domain Shared Objects · 10.1 Canonical Shared Objects · User · Profile · Continu Is this still what you want built? 2.9 min (5) hidden (17 of 19 unknown) — The shell must unify the product through shared objects, not just shared chrome.
11 11. Customer-Facing Product Requirements · 11.1 Shared Customer Shell Requirements · Home Is this still what you want built? 4.5 min (5) 118 / 0 / 0 — The customer-facing shell across web and mobile must provide:
12 11.3 OSHUN Web Requirements · Primary Web Jobs · Web Shell Requirements · Web Tara Require Is this still what you want built? 2.8 min (5) 54 / 0 / 0 — OSHUN Web is the primary breadth, depth, dashboard, and workspace environment.
13 12. Admin-Facing Product Requirements · 12.1 Admin Product Positioning · 12.2 Admin Shared Is this still what you want built? 2.7 min (5) 45 / 1 / 11 — OSHUN Admin is a first-class product, not a collection of internal tools.
14 12.5 Review and Approval Center · 12.6 Trust and Safety Center Is this still what you want built? 2 min (5) hidden (48 of 49 unknown) — This workspace operationalizes the accepted Studio review/approval ADRs as the platform-wide standard.
15 12.7 Content and Editorial Operations · 12.8 Research Integrity and Evidence Operations · Is this still what you want built? 3.2 min (5) hidden (95 of 96 unknown) — Required capabilities:
16 12.14 Incident and Reliability Operations Is this still what you want built? 1.2 min (5) hidden (11 of 11 unknown) — Required capabilities:
17 12.15 Policy and Configuration Center Is this still what you want built? 1.5 min (5) hidden (16 of 16 unknown) — Operators need a dedicated workspace for platform rules, not scattered settings pages.
18 12.16 Executive, BI, and Readiness Views Is this still what you want built? 1.3 min (5) hidden (10 of 10 unknown) — Required capabilities:
19 13. AI Assistant, Teacher, Guru, and Expert System Requirements · 13.1 Product Requirement Is this still what you want built? 1.5 min (5) 15 / 0 / 0 — OSHUN should offer the most compelling, premium, clearly disclosed AI guides and assistants reasonably possibl
20 13.2 Assistant Families · Consumer-Facing Families · Admin-Facing Families Is this still what you want built? 1.4 min (5) hidden (19 of 20 unknown) — The system must support distinct assistant families with shared infrastructure.
21 13.3 Assistant Experience Requirements · 13.4 Disclosure Requirements · 13.5 Persona Regis Is this still what you want built? 2.5 min (5) 60 / 0 / 0 — Required capabilities:
22 13.8 Quality Bar Is this still what you want built? 1.3 min (5) hidden (4 of 5 unknown) — The premium cast should feel unusually high-end, but the design standard is:
23 14. Research, Authoring, and Agentic System Requirements · 14.1 Grounded Research Requirem Is this still what you want built? 1.9 min (5) hidden (24 of 28 unknown) — OSHUN must support serious research, authoring, and evidence-heavy explanation workflows without fragmenting t
24 14.5 Agentic Authoring Requirements Is this still what you want built? 1.2 min (5) 8 / 0 / 0 — Agentic authoring systems must support:
25 14.6 Research Agent Guardrails Is this still what you want built? 1.2 min (5) hidden (6 of 7 unknown) — Agentic research systems must:
26 14.7 Deep Study and Scholarship Requirements Is this still what you want built? 1.2 min (5) 11 / 0 / 0 — The product must support:
27 15. Image, Audio, Video, Avatar, and Model System Requirements · 15.1 Canonical Principle Is this still what you want built? 1.2 min (5) hidden (6 of 6 unknown) — No media-generation capability may be production-critical without:
28 15.2 Isis as the Control Plane Is this still what you want built? 1.3 min (5) hidden (9 of 10 unknown) — All image, audio, video, avatar, and workflow execution must ultimately resolve through an Oshun-controlled co
29 15.3 ComfyUI Requirements · 15.4 RunPod Requirements · 15.5 ElevenLabs and Voice Provider Is this still what you want built? 3.1 min (5) 65 / 0 / 0 — ComfyUI must be treated as:
30 16. Platform Architecture and Canonical Ownership · 16.1 Canonical Platform Split · Iris · Is this still what you want built? 1.9 min (5) hidden (44 of 45 unknown) — The platform ownership model must be explicit and enforced.
31 16.2 Domain Adapter Requirement Is this still what you want built? 1.3 min (5) 11 / 0 / 1 — Every domain and subsystem must connect through explicit adapters rather than ad hoc UI-to-service coupling.
32 16.3 Contract Requirements Is this still what you want built? 1.3 min (5) 14 / 0 / 0 — The platform must define canonical contracts for:
33 16.4 Durability Requirement Is this still what you want built? 1.4 min (5) 7 / 0 / 1 — Any quality-critical runtime using only in-memory state today must be refactored to durable storage, durable q
34 16.5 End-to-End Execution Model · Consumer Assistant Flow · Grounded Research Flow · Media Is this still what you want built? 2.8 min (5) hidden (21 of 22 unknown) — The end-state system should behave predictably across consumer, admin, and media-heavy flows.
35 17. Review, Safety, Compliance, and Governance Requirements · 17.1 Canonical Review Object Is this still what you want built? 1.3 min (5) hidden (10 of 11 unknown) — The accepted review/approval and audit/compliance ADRs must become the platform-wide standard, not a studio-on
36 17.2 Safety Requirements · 17.3 Compliance Requirements · 17.4 Rights and Licensing Requir Is this still what you want built? 3 min (5) 72 / 0 / 1 — The product must support:
37 18. Data, Memory, and Personalization Requirements · 18.1 One Profile, Many Contexts Is this still what you want built? 1.2 min (5) 1 / 0 / 0 — Users must have one unified profile with context-sensitive expressions across domains and assistants.
38 18.2 Personalization Inputs Is this still what you want built? 1.2 min (5) hidden (9 of 9 unknown) — Personalization may incorporate:
39 18.3 Memory Governance · 18.4 Sensitive Memory Controls Is this still what you want built? 1.4 min (5) hidden (10 of 12 unknown) — The product must expose:
40 18.5 Shared Continuity Requirements · 18.6 Data Portability and Recovery Requirements Is this still what you want built? 1.5 min (5) 11 / 0 / 0 — Continuity must span:
41 19. Search, Discovery, Recommendation, and Knowledge Graph Requirements · 19.1 Search Is this still what you want built? 1.2 min (5) hidden (10 of 11 unknown) — Search must unify:
42 19.2 Recommendation Is this still what you want built? 1.3 min (5) hidden (7 of 9 unknown) — Recommendations must support:
43 19.3 Shared Concept Graph Is this still what you want built? 1.3 min (5) hidden (15 of 16 unknown) — The product must maintain a shared concept graph connecting:
44 20. Monetization, Entitlements, and Commerce Requirements · 20.1 Entitlement Principles · Is this still what you want built? 1.4 min (5) hidden (10 of 11 unknown) — Premium should unlock more depth, continuity, guidance, and creation power without undermining trust or core u
45 20.3 Admin Requirements Is this still what you want built? 1.2 min (5) 7 / 0 / 0 — Operators must be able to:
46 21. Analytics, Experimentation, and Quality Requirements · 21.1 Analytics Requirements · 2 Is this still what you want built? 1.6 min (5) 30 / 0 / 0 — The platform must measure:
47 22. Non-Functional Requirements · 22.1 Performance Is this still what you want built? 1.3 min (5) hidden (6 of 7 unknown) — Consumer surfaces must feel fast enough to preserve calm and flow.
48 22.2 Reliability Is this still what you want built? 1.2 min (5) 4 / 0 / 0 — - partial domain outages must degrade gracefully
49 22.3 Security Is this still what you want built? 1.1 min (5) hidden (5 of 5 unknown) — - RBAC and least privilege
50 22.4 Accessibility Is this still what you want built? 1.1 min (5) hidden (6 of 6 unknown) — - WCAG AA minimum for web
51 22.5 Internationalization and Localization Is this still what you want built? 1.1 min (5) hidden (4 of 4 unknown) — - domain content localization support
52 22.6 Privacy Is this still what you want built? 1.1 min (5) hidden (4 of 4 unknown) — - explicit memory controls
53 23. Operational Readiness Requirements · 23.1 Release Readiness Requirements Is this still what you want built? 1.6 min (5) 23 / 0 / 0 — The product is not complete until it is operable at scale.
54 24. Success Metrics · 24.1 Consumer Metrics · 24.2 Trust and Quality Metrics · 24.3 Platfo Is this still what you want built? 2.1 min (5) hidden (26 of 27 unknown) — The product must be evaluated as one integrated system.
55 25. Phased Delivery Model · Phase 0: Alignment · Phase 1: Shared Contract Expansion · Phas Is this still what you want built? 2.2 min (5) hidden (27 of 28 unknown) — The system should be delivered in phases, but the end-state architecture must be designed from the start.
56 26. Anti-Patterns to Avoid Is this still what you want built? 1.5 min (5) hidden (10 of 10 unknown) — - adding Nisaba as just another icon
57 27. Acceptance Criteria for the End-State V1 · 27.1 Product Cohesion · 27.2 Consumer Featu Is this still what you want built? 2.8 min (5) hidden (27 of 28 unknown) — OSHUN V1 is complete only when all of the following are true.
58 Closing Position Is this still what you want built? 1.5 min (5) hidden (8 of 9 unknown) — The fully built OSHUN product should not be understood as "a meditation app with extra tabs" or "an admin dash

ADR-0004: Eventing Strategy - Event Broker and Contract Patterns#

Revision 1 · adr reviewed as adr · draft · 9 chunks · about 33 minutes in all (revision 12: 9 chunks, about 33 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 Context and Problem Statement Do you accept this decision and its consequences? 2 min (5) hidden (15 of 17 unknown) 1: metadata header The Oshun monorepo requires a unified eventing strategy to enable asynchronous communication between services,
2 Decision Drivers Do you accept this decision and its consequences? 1.4 min (5) hidden (8 of 9 unknown) — - **Scalability**: Support 100k+ concurrent users with high-throughput events
3 Considered Options · Event Broker Options · Option A: Redis (Pub/Sub + Streams) + BullMQ ( Do you accept this decision and its consequences? 4 min (5) hidden (72 of 75 unknown) — **Description**: Use Redis as the primary message broker with Pub/Sub for real-time fan-out events, Redis Stre
4 Decision Outcome · Event Broker Decision · Event Contract Decision Do you accept this decision and its consequences? 3.1 min (5) hidden (22 of 25 unknown) — **Chosen Event Broker**: Option A - Redis (Pub/Sub + Streams) + BullMQ **Chosen Contract Pattern**: Contract O
5 Implementation Details · Technical Specifications · Event Flow Architecture · Event Naming Do you accept this decision and its consequences? 14.8 min (15) hidden (15 of 21 unknown) — **Event Envelope Schema**:
6 Consequences · Positive Consequences · Negative Consequences · Risks and Mitigation Do you accept this decision and its consequences? 3.4 min (5) hidden (9 of 12 unknown) — - ✅ **Unified Eventing**: Single pattern for all async communication
7 Compliance and Security · Security Implications · Compliance Requirements Do you accept this decision and its consequences? 1.4 min (5) 4 / 0 / 3 — - **Event Encryption**: Sensitive payloads encrypted at rest
8 Monitoring and Observability · Metrics to Track · Alerting Strategy Do you accept this decision and its consequences? 1.4 min (5) hidden (7 of 7 unknown) — - **Publishing**: Events published/second, publish latency, publish errors
9 Related Decisions · Upstream Dependencies · Downstream Impacts Anything wrong or missing here? 1.3 min (5) hidden (5 of 6 unknown) 12: boilerplate: References; boilerplate: Revision History - **ADR-0001**: Git Consolidation (single repo for shared event packages)

ADR-0005: API Contract Approach - OpenAPI and Protocol Buffers#

Revision 1 · adr reviewed as adr · draft · 9 chunks · about 31 minutes in all (revision 12: 9 chunks, about 31 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 Context and Problem Statement Do you accept this decision and its consequences? 2 min (5) hidden (15 of 18 unknown) 1: metadata header The Oshun monorepo requires a standardized approach to API contracts that enables type-safe communication betw
2 Decision Drivers Do you accept this decision and its consequences? 1.4 min (5) hidden (8 of 9 unknown) — - **External Client Support**: Web browsers, mobile apps, third-party integrations
3 Considered Options · Option 1: OpenAPI Only · Option 2: Protocol Buffers Only · Option 3: Do you accept this decision and its consequences? 4 min (5) hidden (36 of 40 unknown) — **Description**: Standardize on OpenAPI 3.1 for all API contracts. Use REST/HTTP for all service communication
4 Decision Outcome Do you accept this decision and its consequences? 2.9 min (5) hidden (21 of 23 unknown) — **Chosen option**: Option 3 - Both OpenAPI and Protocol Buffers
5 Implementation Details · Technical Specifications · API Design Guidelines · Migration Stra Do you accept this decision and its consequences? 13.6 min (15) hidden (35 of 42 unknown) — **OpenAPI Directory Structure**:
6 Consequences · Positive Consequences · Negative Consequences · Risks and Mitigation Do you accept this decision and its consequences? 3.4 min (5) hidden (11 of 12 unknown) — - ✅ **Optimized for Use Case**: REST for external, gRPC for internal
7 Compliance and Security · Security Implications · Compliance Requirements Do you accept this decision and its consequences? 1.4 min (5) hidden (4 of 7 unknown) — - **Input Validation**: All inputs validated against schemas
8 Monitoring and Observability · Metrics to Track · Alerting Strategy Do you accept this decision and its consequences? 1.3 min (5) hidden (6 of 6 unknown) — - **OpenAPI**: Request count, latency, error rates per endpoint
9 Related Decisions · Upstream Dependencies · Downstream Impacts Anything wrong or missing here? 1.3 min (5) hidden (7 of 7 unknown) 12: boilerplate: References; boilerplate: Revision History - **ADR-0001**: Git Consolidation (single repo for contract packages)

Oshun Monorepo — Architecture Overview#

Revision 1 · architecture reviewed as architecture · draft · 9 chunks · about 18 minutes in all (revision 12: 9 chunks, about 18 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 Oshun Monorepo — Architecture Overview Is this boundary or contract correct? 1.4 min (5) 0 / 1 / 0 — Rewritten 2026-07-16 from the measured state of the repository (audit R-21, 'docs/audits/MONOREPO_ARCHITECTURE
2 The shape of the system Is this boundary or contract correct? 2.4 min (5) hidden (3 of 5 unknown) — Oshun is one polyglot monorepo carrying **ten products (V1–V10)** on a shared platform of **48 domain librarie
3 Domain Responsibilities Matrix Is the rule that generates this table right? 2.8 min (5) hidden (2 of 3 unknown) 48: generated rows, checked by node tools/domains/check-registry.mjs Generated from 'domains.json' (run 'node tools/domains/check-registry.mjs' to verify it is current). Projects
4 Data Ownership Matrix Is this boundary or contract correct? 3.3 min (5) hidden (3 of 3 unknown) — Domain isolation extends to storage: every domain owns its database and no domain reads another domain's table
5 Event Flow Is this boundary or contract correct? 1.7 min (5) hidden (3 of 5 unknown) — Cross-domain integration is event-driven where it is asynchronous and API-mediated where it is synchronous:
6 API Gateway & BFF layer Is this boundary or contract correct? 1.5 min (5) 0 / 1 / 0 — North-south traffic enters through per-product BFFs ('apps/oshun/bff' is the richest: auth, entitlements, idem
7 Build system Is this boundary or contract correct? 1.8 min (5) hidden (3 of 5 unknown) — - **Nx** drives ~3,400 projects; affected-based CI ('ci.yml') runs lint/typecheck/build/test against merge-bas
8 Infrastructure Is this boundary or contract correct? 1.3 min (5) 0 / 1 / 0 — Three live IaC roots are mapped authoritatively in 'infra/README.md': 'infra/terraform-v1' (canonical V1 ECS F
9 Where to look next Anything wrong or missing here? 1.6 min (5) hidden (3 of 5 unknown) — - 'domains.json' — machine-readable domain registry (R-19)

OSHUN V1 Persistence Migration and Backfill Plan#

Revision 1 · spec reviewed as spec · draft · 14 chunks · about 46 minutes in all (revision 12: 14 chunks, about 46 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 OSHUN V1 Persistence Migration and Backfill Plan Is this boundary or contract correct? 1.2 min (5) 0 / 1 / 0 — This runbook is generated from 'libs/oshun/persistence/src/migration-plan.ts'. Every domain plan is dry-runnab
2 Shared Execution Contract Is this boundary or contract correct? 1.4 min (5) 5 / 1 / 0 — - Target schema: 'libs/oshun/persistence/prisma/schema.prisma'
3 tara Is this boundary or contract correct? 3.6 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-tara'
4 arete Is this boundary or contract correct? 3.7 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-arete'
5 veritas Is this boundary or contract correct? 3.7 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-veritas'
6 nyx Is this boundary or contract correct? 3.5 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-nyx'
7 nisaba Is this boundary or contract correct? 3.8 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-nisaba'
8 metis Is this boundary or contract correct? 3.8 min (5) 11 / 2 / 0 — - Migration key: 'v1-contract-persistence-metis'
9 library Is this boundary or contract correct? 3.4 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-library'
10 atelier Is this boundary or contract correct? 3.4 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-atelier'
11 studio Is this boundary or contract correct? 3.4 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-studio'
12 messaging Is this boundary or contract correct? 3.4 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-messaging'
13 identity Is this boundary or contract correct? 3.4 min (5) 10 / 3 / 0 — - Migration key: 'v1-contract-persistence-identity'
14 cross-cutting Is this boundary or contract correct? 4.1 min (5) 11 / 2 / 0 — - Migration key: 'v1-contract-persistence-cross-cutting'

Sophia Domain — Technical Specifications#

Revision 1 · guide reviewed as technical-spec · draft · 30 chunks · about 100 minutes in all (revision 12: 30 chunks, about 100 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 Sophia Domain — Technical Specifications Is this boundary or contract correct? 1.3 min (5) 0 / 1 / 1 — > Knowledge engine, RAG and semantic-search domain of the Oshun monorepo.
2 Two Contract Layers — Understanding the Distinction Is this boundary or contract correct? 1.9 min (5) 1 / 2 / 1 — Sophia carries **two distinct, independently-authored contract layers**, and the distinction matters throughou
3 1. Persistence Layer — Prisma Schema Is this boundary or contract correct? 4.1 min (5) 0 / 2 / 2 — **File:** 'libs/sophia/database/prisma/schema.prisma' · **Generator output:** 'libs/sophia/database/src/genera
4 1.1 Document (documents) Is this boundary or contract correct? 4.8 min (5) hidden (5 of 5 unknown) — The 'Document' model is the primary record for every piece of knowledge in Sophia. It stores full bibliographi
5 1.2 DocumentChunk (document_chunks) Is this boundary or contract correct? 3.5 min (5) hidden (3 of 4 unknown) — A 'DocumentChunk' is one segment of a document after the chunking stage has been applied. Each chunk stores it
6 1.3 Citation (citations) Is this boundary or contract correct? 3.5 min (5) hidden (3 of 3 unknown) — A 'Citation' record links a claim in some consumer context to the specific source document (and optionally chu
7 1.4 Entity (entities) — Knowledge-Graph Node Is this boundary or contract correct? 3.7 min (5) hidden (4 of 4 unknown) — An 'Entity' is a node in the knowledge graph. It represents a named real-world thing (a person, place, concept
8 1.5 EntityMention (entity_mentions) Is this boundary or contract correct? 3.1 min (5) 0 / 1 / 1 — An 'EntityMention' records a specific occurrence of an entity within a document. This is what allows Sophia to
9 1.6 Relation (relations) — Knowledge-Graph Edge Is this boundary or contract correct? 3.3 min (5) hidden (2 of 2 unknown) — A 'Relation' is a directed, typed edge between two 'Entity' records. The unique constraint '[sourceId, targetI
10 1.7 Knowledge Packs Is this boundary or contract correct? 1.8 min (5) hidden (5 of 5 unknown) — Knowledge packs are curated, versioned collections of documents and entities. The 'KnowledgePack' model ('know
11 1.8 Canonical Grounding Contracts Is this boundary or contract correct? 2.7 min (5) hidden (9 of 10 unknown) — Eight models form the grounding contract layer that was added on top of the core research data. These models r
12 1.9 IngestionJob and AuditLog Is this boundary or contract correct? 1.8 min (5) hidden (3 of 3 unknown) — 'IngestionJob' ('ingestion_jobs') tracks the state of a document as it moves through the ingestion pipeline. E
13 1.10 Prisma Enums Is this boundary or contract correct? 4.1 min (5) hidden (3 of 3 unknown) — All Prisma enum values are uppercase strings. The table below gives each enum's name, member count, and comple
14 2. Validation Contract Layer — @sophia/schemas Is this boundary or contract correct? 1.4 min (5) 0 / 1 / 0 — **Directory:** 'libs/sophia/schemas/src/'. All schemas are Zod schemas exported from 'index.ts'. Enum members
15 2.1 Branded ID Types (ids.ts) Is this boundary or contract correct? 1.6 min (5) hidden (4 of 4 unknown) — Sophia uses branded ID types to prevent accidentally passing a 'CitationId' where a 'DocumentId' is expected.
16 2.2 Research Document Schemas (research-document.ts) Is this boundary or contract correct? 3.8 min (5) hidden (5 of 5 unknown) — 'ResearchDocumentSchema' is the main document object at the validation layer. It is composed from several comp
17 2.3 Citation Schemas (citation.ts) Is this boundary or contract correct? 4.3 min (5) hidden (6 of 7 unknown) — The citation schema file defines both the bibliographic source model and the citation metadata model. The key
18 2.4 Claim Schemas (claim.ts) · Claim Origin Labelling Is this boundary or contract correct? 4.2 min (5) hidden (9 of 10 unknown) — The claim schema file defines types for classifying and tracking individual claims within AI-generated text. T
19 2.5 Knowledge Pack Schemas (knowledge-pack.ts) Is this boundary or contract correct? 3.3 min (5) hidden (3 of 3 unknown) — The knowledge pack Zod enums differ from their Prisma counterparts in both member count and values. Notably, t
20 2.6 Entity Schemas (entity.ts) — Knowledge-Graph Nodes Is this boundary or contract correct? 3.4 min (5) hidden (3 of 3 unknown) — Note the casing difference for 'EntityType' between the two layers: the Zod enum uses title-cased strings (''T
21 2.7 Relation Schemas (relation.ts) — Knowledge-Graph Edges Is this boundary or contract correct? 3.9 min (5) hidden (10 of 11 unknown) — The 'RelationTypeSchema' Zod enum has **57 members** — significantly more than the Prisma 'RelationType' (27 m
22 2.8 Sophia Error Types (errors.ts) Is this boundary or contract correct? 1.8 min (5) hidden (11 of 11 unknown) — Sophia defines a structured error hierarchy that allows callers to distinguish error types programmatically —
23 3. Applications · 3.1 Search API (apps/sophia/search-api) · 3.2 Knowledge Graph (apps/soph Is this boundary or contract correct? 12 min (15) hidden (35 of 38 unknown) — Four applications live under 'apps/sophia/'. The two backend services ('search-api', 'knowledge-graph') run a
24 4. Events Is this boundary or contract correct? 4.7 min (5) hidden (5 of 6 unknown) — All Sophia domain events are defined in 'libs/contracts/src/events/sophia.ts' and published through '@sophia/e
25 5. Vector Store Layer — @sophia/vectordb · 5.1 Predefined Collection Schemas Is this boundary or contract correct? 5.3 min (10) hidden (5 of 6 unknown) — '@sophia/vectordb' provides a unified vector-store abstraction over four backends, each implementing a common
26 6. Indexing Layer — @sophia/indexing · 6.1 Chunking Strategies · 6.2 Embedding Providers Is this boundary or contract correct? 1.8 min (5) 0 / 3 / 2 — 'indexing/src/chunking/strategies.ts' implements **8 chunker classes**, all extending 'BaseChunker'. The 'Chun
27 7. Configuration and Environment Variables Is this boundary or contract correct? 3.3 min (5) hidden (3 of 3 unknown) — The table below lists all environment variables consumed by Sophia's applications and libraries. Services bind
28 8. V2 Release-Candidate Ingestion Contract Is this boundary or contract correct? 1.8 min (5) 1 / 1 / 1 — '@sophia/ingestion' is the source-of-truth ingestion pipeline for all production Sophia usage. The V2 release-
29 9. Cross-Domain Integration Is this boundary or contract correct? 1.7 min (5) hidden (9 of 9 unknown) — Sophia integrates with other domains through a small, well-defined set of channels. It depends on shared infra
30 10. Acceptance Criteria Is this boundary or contract correct? 2 min (5) hidden (9 of 9 unknown) — A Sophia change is acceptance-complete when all of the following hold:

V1 Specification Coverage Matrix (2026-05-30)#

Revision 1 · spec reviewed as spec · draft · 6 chunks · about 17 minutes in all (revision 12: 6 chunks, about 17 minutes) · 0 over a 15-minute session

# Chunk Question Estimate Claims: proposed / existing / unknown Not asked Opens with
1 V1 Specification Coverage Matrix (2026-05-30) Is this boundary or contract correct? 2.2 min (5) hidden (2 of 2 unknown) — Systematic, spec-by-spec analysis of the **entire** 'V1/features.md' (6,241 lines: 22 top-level sections, 99 s
2 A. features.md section coverage (22 sections) Is this boundary or contract correct? 4.7 min (5) 0 / 1 / 0 — / § / Section / Status / Evidence / note / / -------------------------------------------- / ------------------
3 B. Customer domains (6) — Exit-Criterion "one coherent product" Is this boundary or contract correct? 3 min (5) hidden (1 of 1 unknown) — / Domain / Status / Evidence / / --------------------- / ------------- / -------------------------------------
4 C. V1 Exit Criteria scorecard (the spec's definition of done — 19) Is this boundary or contract correct? 3.9 min (5) 0 / 1 / 0 — / # / Criterion / Verdict / / --- / -------------------------------------------------------------- / ---------
5 D. Consolidated gap register (all layers, not just studio) Is this boundary or contract correct? 1.8 min (5) 1 / 1 / 2 — 1. **Operator/creator Studio pages (~450 unwired of 469)** — the dominant gap. Backends/algorithms exist; the
6 E. Verdict Is this boundary or contract correct? 1.7 min (5) hidden (3 of 3 unknown) — - **Phase 1 (analyze entire implementation):** done here — all 22 spec sections, 6 domains, 19 Exit Criteria w

The golden plans in tools/eve-content/testing/fixtures/granular-review-planner are regenerated under egr-planner-4, and EGR.18/EGR.19 fixtures will use it.