Eve library

OSHUN V1 Program Charter

Spec

Revision 1

On this page

Status: Accepted v1.3
Date: March 23, 2026
Owners: OSHUN Product, OSHUN Engineering, OSHUN Design, OSHUN Safety, OSHUN Platform
Related:

  • docs/releases/v1/reports/unified-product-gap-analysis-and-nisaba-integration-report-2026-03-22.md
  • docs/releases/v1/reports/metis-domain-gap-analysis-and-full-integration-report-2026-03-23.md
  • docs/releases/v1/reports/metis-v1-workstream-tracking.md
  • docs/releases/v1/reports/subsystem-ownership-matrix.md
  • docs/releases/v1/specs/domain-service-consolidation-strategy.md
  • docs/releases/v1/specs/public-web-and-app-distribution-tracking.md
  • docs/releases/v1/specs/web-app-admin-v1-prd.md
  • docs/releases/v1/specs/product-brief.md
  • docs/releases/v1/scope/mvp-scope.md
  • docs/releases/v1/design/ux-principles.md
  • docs/releases/v1/specs/todos.md

1. Purpose and Authority#

This document is the canonical governance charter for OSHUN V1.

Its job is to turn the March 22, 2026 gap analysis and the unified V1 PRD into one enforceable program model for planning, execution, release gating, and cross-functional decision-making.

For OSHUN V1 end-state decisions, this charter takes precedence over earlier MVP-era framing that described OSHUN as a four-domain shell and over March 22, 2026 planning snapshots that predated the Metis corrective scope. Earlier MVP documents remain useful historical references for codebase reality and launch constraints, but they are not the source of truth for the final V1 product shape.

2. Charter Mandates#

The program is governed by the following mandates:

  1. OSHUN is one contemplative intelligence product, not a loose bundle of apps.
  2. Tara is the emotional and experiential center of gravity for the product.
  3. Nisaba is a first-class Oshun domain, not an optional add-on.
  4. Metis is a first-class Oshun domain and the shared educational substrate for course, tutor, assessment, and cross-domain learning journeys.
  5. Sophia is the grounding, citation, and research-integrity substrate behind relevant experiences, not a naive shell peer.
  6. Admin web and admin mobile are first-class products required for launch quality, governance, and operations.
  7. Iris, Psyche, Lilith, Sophia, and Isis have explicit ownership boundaries and must not collapse into an ambiguous assistant blob.
  8. Quality, safety, accessibility, observability, provenance, privacy, and review are launch gates, not polish.

3. Canonical Product Model#

3.1 Customer-Facing Domain Model#

Surface Canonical Role in V1 Primary Value
Tara experiential spine centering, rituals, transitions, sleep, presence
Arete disciplined embodiment habits, goals, reflection, daily structure
Veritas epistemic clarity grounded claims, briefings, evidence inspection
Nyx wonder and perspective sky awareness, events, cosmological orientation
Nisaba textual depth and lineage passages, study, compare, annotation, scholarship
Metis educational substrate courses, tutoring, assessments, study programs, learning transfer
Sophia substrate, not shell peer grounding, citations, notebooks, evidence, research integrity

Metis is both a peer Oshun domain in the V1 product model and a shared educational substrate that can turn content, questions, or goals from other domains into governed learning experiences. That does not move source-of-record ownership away from the originating domain or the core substrates that own grounding, memory, pedagogy policy, or generation control.

3.2 Product Surfaces#

Surface Role
Customer Web deep research, reading, study, dashboards, multi-panel work
Customer Mobile routine entry, quick actions, continuity, notifications, daily return
Customer PWA installable web continuation path
Admin Web canonical operator cockpit for review, governance, support, privacy, model/persona ops
Admin Mobile urgent review, escalation, acknowledgements, on-call companion

3.3 Canonical Assistant and Platform Ownership#

Subsystem Owns Must Not Be Treated As
Iris persistent memory, assistant state, privacy-bounded recall, consent-aware memory controls the real-time avatar/runtime layer
Psyche real-time multimodal text/voice/avatar session orchestration the durable memory or grounding source of record
Lilith contemplative teacher logic, tone policy, persona safety, domain-specific spiritual safeguards the shell router or generic product orchestration layer
Sophia grounding, retrieval, citations, evidence packs, notebooks, knowledge graph, research integrity a fifth shell tab competing with Tara for product center
Isis generation control plane, model/workflow registry, provenance, promotion and release governance an ad hoc direct-runtime bypass layer

4. Alignment to the Gap Analysis and PRD#

Gap Analysis / PRD Signal Charter Directive
Shell currently behaves like four domains in one container Expand the governing model to Tara, Arete, Veritas, Nyx, Nisaba, and Metis, with Sophia behind the scenes
Tara is strong but not the product's center Make Tara the canonical emotional center and framing layer for Oshun
Web and mobile maturity differ Treat cohesion, role clarity, and cross-platform fit as program requirements, not optional cleanup
Nisaba is strategically high leverage Make Nisaba a first-class domain with web-deep and mobile-companion roles
Metis was omitted from the March 22 scope model Add Metis as a first-class domain and educational substrate across shell, contracts, admin, and governance
Sophia should not be a naive shell peer Define Sophia as the grounding substrate for Veritas, Nisaba, assistant, and operator flows
No unified operator product exists today Treat admin web and admin mobile as launch-blocking first-class products
Assistant ownership is blurred Enforce explicit Iris/Psyche/Lilith/Sophia/Isis ownership boundaries
Safety/review exists but lacks one operating model Require one cross-product governance, review, and release-gate model

5. Scope and Program Boundaries#

This charter governs:

  • customer web, customer mobile, and PWA
  • admin web and admin mobile
  • shared navigation, auth, search, personalization, assistant UX, and BFF
  • domain adapters and cross-domain object models
  • Metis learner, tutor, course, assessment, and cross-domain educational orchestration surfaces
  • grounding, memory, persona, voice/avatar, and generation-control substrates
  • moderation, privacy, rights, provenance, review, compliance, and release governance
  • launch readiness, scorecards, and operating cadences

This charter does not replace:

  • implementation tickets and engineering task decomposition
  • policy/legal source documents
  • team staffing plans or budget decisions

6. Master Dependency Map#

Metis must be tracked in this dependency map as its own launch workstream, not only as a row inside the domain layer. The canonical Metis-specific tracking artifact is docs/releases/v1/reports/metis-v1-workstream-tracking.md.

6.1 Layered Dependency Model#

Layer Primary Elements Depends On
Experience Surfaces customer web, customer mobile, PWA, admin web, admin mobile shell orchestration, BFF, shared UI, domain adapters
Shell Orchestration navigation, auth, domain registry, analytics, notifications, assistant UX, feature flags shared contracts, BFF, platform services
Domain Experience Layer Tara, Arete, Veritas, Nyx, Nisaba, Metis surfaces and adapters BFF/domain APIs, shared models, Sophia where grounded
Core Substrates Iris, Psyche, Lilith, Sophia, Isis contracts, storage, observability, policy systems
Shared Platform Services identity, entitlements, search, offline/cache, observability, audit, privacy, experiment, notification infrastructure persistence, infra, contracts
Governance and Operations moderation, review, privacy, rights, support, incident, release readiness admin surfaces, audit, telemetry, policy, platform health

6.2 Critical Cross-Functional Dependency Edges#

Dependency Edge Why It Matters
Customer shell -> BFF -> domain adapters core product continuity and route coherence
Veritas/Nisaba/assistant -> Sophia grounded claims, citations, and research integrity
Tara/Arete/Veritas/Nyx/Nisaba -> Metis cross-domain materials and intent must be able to become governed learning experiences without ownership drift
Metis -> Sophia/Iris/Lilith/Isis/Psyche/Themis grounded tutoring, memory, pedagogy, generation control, embodiment, and integrity controls must remain explicit
Assistant UX -> Iris/Psyche/Lilith/Sophia memory, embodiment, tone, and grounding must stay separable and composable
Admin web/mobile -> audit/review/policy services governance work cannot be delegated to ad hoc tooling
Standalone domain apps -> BFF/domain adapters shell auth, routing, and contract boundaries must not fragment into direct client-to-domain integrations
Isis -> model/workflow providers generation must be promoted through controlled contracts
Shared contracts -> all surfaces six-domain navigation, object models, analytics, and permissions must remain versioned and coherent

7. Cross-Functional Milestone Plan#

Metis milestone participation and evidence expectations are tracked in docs/releases/v1/reports/metis-v1-workstream-tracking.md.

Milestone Goal Primary Workstreams Exit Condition
M0 Program Alignment unify product framing and governance posture product, design, engineering, safety, platform charter, scorecards, release model, ownership, and dependency map accepted
M1 Shell Foundations establish six-domain shell and shared contracts web, mobile, BFF, shared libs shell contracts, routing, auth, analytics, notifications, and domain registry recognize final model
M2 Customer Cohesion ship coherent customer shell and high-value journeys design, web, mobile, content Tara-led home, cross-domain continuity, search, library, profile, and critical entry flows verified
M3 Domain Depth and Grounding integrate Nisaba, Metis, and grounding-rich flows Nisaba, Veritas, Metis, Sophia, assistant reading, evidence, annotations, grounded explainers, learning flows, and educational handoffs verified
M4 Operator Readiness ship admin web/mobile critical workflows admin, safety, privacy, support, model/persona ops operator-critical queues, review flows, privacy/rights tooling, and incident workflows verified
M5 Launch Readiness prove customer, admin, platform, governance, and Metis quality QA, SRE, product, support, compliance, Metis ops scorecards green, Metis overlay reviewed, release gates passed, blockers resolved or exception-approved

8. Launch Blockers vs. Phase-Later Depth#

8.1 Hard Launch Blockers#

  • Tara-centered customer framing in the shell
  • Nisaba present in the canonical shell/domain model
  • Metis present in the canonical shell/domain model and governed as the shared educational substrate for course, tutor, and assessment flows
  • Sophia integrated as grounding substrate for grounded customer and operator flows
  • customer web/mobile critical journeys implemented and automated
  • admin web critical review/governance/support/privacy flows implemented and automated
  • admin mobile urgent queue and escalation baseline implemented and automated
  • durable state for review, approval, memory boundaries, provenance, and generation workflows
  • release readiness scorecards and gates functioning
  • Metis launch scorecard and blocker lane visible in the V1 readiness dashboard
  • standalone domain app consolidation strategy accepted for shell-serving backends and superseded frontends
  • public website, store distribution, and external deep-link workstreams tracked as launch blockers
  • accessibility, performance, privacy, safety, and observability gates passing

8.2 Phase-Later Depth#

  • long-tail localization beyond the launch set
  • non-critical creative/experimental assistant surfaces
  • deeper collaboration and authoring features not required for launch safety or product coherence
  • extended content breadth that does not affect launch-critical loops

9. Release Gates#

Metis channel-specific gate requirements and first-class launch-lane tracking are operationalized in docs/releases/v1/reports/metis-v1-workstream-tracking.md.

Channel Required Gate Outcome
Alpha architecture, contracts, auth, routing, telemetry, and critical shell flows exist with owner-assigned open risks
Private Beta customer critical journeys, admin critical journeys, grounding baseline, privacy baseline, and automation baseline pass with blocker tracking
Public Beta retention-critical flows, trust signals, accessibility, performance, and support/on-call readiness pass at beta thresholds
General Availability customer/admin/platform/governance scorecards are green; Metis overlay is green or exception-approved; public website, store listings, and deep-link distribution are ready; no unresolved launch-blocking issues remain without approved, time-bounded exceptions

9.1 Gate Rules#

  • No channel promotion without an explicit scorecard snapshot.
  • No GA promotion while a red launch blocker remains open.
  • No GA promotion while the Metis overlay is absent for releases that include educational or tutoring scope.
  • Exceptions must be written, owner-assigned, time-bounded, and approval-backed.
  • Automated verification evidence must be attached to the relevant gate review.

10. Canonical Status Taxonomy#

Status Meaning
proposed idea exists but scope is not yet accepted
defined scope, owner, and acceptance criteria are written
ready dependencies are satisfied and work can begin
in_progress implementation or documentation is underway
blocked work cannot proceed because a dependency or decision is unresolved
in_review code, design, or governance review is active
verified applicable automation and evidence are passing
done implementation, verification, and required signoff are complete
deferred intentionally moved out of current release scope
superseded replaced by a newer approved task or decision
canceled explicitly abandoned with rationale

11. Owner Roles and Signoff Model#

Role Primary Responsibility Required Signoff Areas
Product Lead scope, priorities, release decisions product acceptance, milestone readiness
Design Lead IA, interaction quality, visual cohesion, ergonomics UX acceptance, design exceptions
Engineering Lead implementation feasibility and delivery integrity technical acceptance, delivery risk
Platform Lead contracts, durability, observability, rollout safety platform readiness, rollback posture
Safety Lead trust/safety policy, abuse review, disclosure posture safety gates, high-risk flows
Compliance/Privacy Lead privacy, rights, governance, audit posture privacy, rights, legal-compliance gates
Editorial/Content Lead customer-facing content, teacher/curriculum integrity content, instructional, and publishing gates
Support/Ops Lead support readiness, incident readiness, operator workflows support, launch operations, incident readiness

12. Weekly Operating Cadences#

12.1 Cross-Domain Architecture Review#

  • Frequency: weekly
  • Default focus: contracts, dependency risks, durability gaps, platform ownership, rollout safety
  • Required attendees: engineering lead, platform lead, domain leads, BFF/shared contracts owners

12.2 Product Quality Review#

  • Frequency: weekly
  • Default focus: customer journeys, shell coherence, ergonomics, accessibility, performance, automation posture
  • Required attendees: product lead, design lead, engineering lead, QA owner

12.3 Trust/Safety/Compliance Review#

  • Frequency: weekly
  • Default focus: disclosure, moderation, privacy, rights, review queues, approval evidence, policy exceptions
  • Required attendees: safety lead, compliance/privacy lead, product lead, operator lead

13. Weekly Launch-Readiness Dashboard#

The weekly launch-readiness dashboard must expose:

  • customer scorecard status
  • admin scorecard status
  • platform scorecard status
  • governance scorecard status
  • top blocking risks and owners
  • automation pass/fail summaries
  • release gate state by channel
  • unresolved exception count and expiry
  • documentation remediation status
  • support/on-call readiness status

14. Canonical Readiness Scorecards#

14.1 Customer-Facing Complete#

Required dimensions:

  • Tara-led shell coherence
  • six-domain navigation and continuity
  • Metis learner entry and cross-domain educational handoffs
  • critical web/mobile/PWA flows
  • grounded trust signals where required
  • accessibility, performance, and resilience
  • Playwright/mobile E2E coverage for critical journeys

14.2 Admin-Facing Complete#

Required dimensions:

  • admin web critical queues and detail workflows
  • admin mobile urgent review and escalation flows
  • RBAC and audit continuity
  • operator ergonomics and keyboard-first operation where applicable
  • privacy, support, incident, and policy workflows

14.3 Platform Complete#

Required dimensions:

  • contract/version integrity
  • durable state and idempotency
  • observability and alert posture
  • rollout/rollback readiness
  • build, test, and dependency health

14.4 Governance Complete#

Required dimensions:

  • release gates and scorecard integrity
  • decision logging and exception traceability
  • privacy, rights, safety, and review controls
  • support/on-call preparedness
  • documentation and runbook currency

15. Risk Register#

Risk ID Risk Impact Mitigation Owner
R-001 Tara framing remains cosmetic instead of structural incoherent product identity require Tara-led shell/home/continuity acceptance criteria Product + Design
R-002 Nisaba is added as a fifth icon without meaningful role clarity weak adoption and IA sprawl require web-deep/mobile-companion scope and study workflows Product + Nisaba
R-003 Sophia is surfaced as a competing shell peer confused IA and duplicated research UX enforce Sophia-as-substrate rule in contracts and UX reviews Product + Sophia
R-004 Admin product remains fragmented unsafe operations and delayed launches treat admin web/mobile as launch blockers with dedicated scorecards Product + Ops
R-005 Assistant ownership remains blurred privacy, tone, and grounding regressions adopt explicit Iris/Psyche/Lilith/Sophia/Isis matrix Platform
R-006 Automation lags behind UI and governance changes false readiness and regression risk require automation as part of feature completion QA + Engineering
R-007 Review, approval, and provenance state remain non-durable audit and rollback failures require durable persistence and replayability Platform
R-008 Scope expands without blocker discipline missed launch windows use blocker-vs-phase-later rubric and weekly triage Product
R-009 Documentation drift reintroduces four-domain framing planning confusion and execution churn treat charter as canonical and maintain remediation queue Product Ops
R-010 Cross-domain dependencies are hidden until late integration launch instability maintain dependency map and weekly architecture review Engineering + Platform
R-011 Metis is treated as either only a shell tab or only a backend broken educational role clarity enforce explicit Metis-as-domain-and-substrate boundaries Product + Metis

16. Backlog Hygiene Rule#

  • No task may enter implementation without scope, owner, dependencies, and acceptance criteria.
  • Any task larger than roughly 1-3 focused working days must be split before implementation.
  • Tasks missing applicable automation must either include that automation work or be blocked by an explicit prerequisite.
  • Blocked tasks must name the blocker owner and required unblock condition.

17. PRD Exception Decision Log#

17.1 Decision Log Rules#

  • Every exception to the PRD or this charter must be logged before release.
  • Each exception must record rationale, owner, risk, duration, and review date.
  • No open-ended exception is valid for GA.

17.2 Initial Governing Decisions#

Decision ID Decision Status
DEC-001 Tara-centered framing is the canonical Oshun V1 posture accepted
DEC-002 Nisaba is a first-class Oshun domain accepted
DEC-003 Sophia is the grounding substrate, not a shell peer accepted
DEC-004 Admin web and admin mobile are launch-critical products accepted
DEC-005 Iris/Psyche/Lilith/Sophia/Isis ownership is explicit and non-overlapping by default accepted
DEC-006 Metis is a first-class Oshun domain and the shared educational substrate accepted
DEC-007 Standalone domain apps serve the shell only through the BFF and typed adapters accepted
DEC-008 Public website and mobile-store distribution are launch-blocking workstreams accepted

18. Documentation Contradiction Audit Seed#

This seed audit identifies high-signal top-level contradictions that must not drive V1 decisions.

Document Current Contradiction Resolution in This Charter
docs/releases/v1/specs/product-brief.md frames Oshun as four domains and a four-surface shell model treat as MVP baseline; defer canonical V1 framing to this charter and the unified PRD
docs/releases/v1/scope/mvp-scope.md defines universal search and IA around four domains only treat as historical MVP boundary, not end-state V1 direction
docs/releases/v1/specs/readme.md points to MVP-era product framing without a V1 charter add charter and PRD as canonical V1 references
docs/releases/v1/design/ux-principles.md shared-pattern language omits Nisaba update principles to reflect the final six-domain shell
March 22 charter snapshot encoded a pre-Metis V1 model accept the March 23 Metis corrective scope into the canonical charter

Tasks V1-GOV-024 and V1-GOV-025 remain open until the broader remediation pass is completed across all affected Oshun documentation.

19. Acceptance and Task Mapping#

This charter is intended to close the following backlog items by providing one canonical source of truth:

  • V1-GOV-001 through V1-GOV-023
  • V1-GOV-026

The following items are only seeded here and remain open:

  • V1-GOV-024
  • V1-GOV-025