# Eve — Project operating system: owner workflow, outcomes and evidence

## Decision and status

On 9 October 2026 Elikem accepted the recommendation to make Eve a project-centred operating system, with a daily **Needs you** home and tasks/families underneath. His reply was: “yes lets go with your recommendation”. This records the accepted direction and authorization to prepare the proposal and ordered backlog. This new PRD revision still requires the normal independent review and product approval; it is not self-approved and does not authorize changed privileges, automation switches or publication.

First pilot: **Eve’s own owner workflow**. Demonstrate observed need → evidence → proposal → Elikem’s decision → existing execution → independent verification → observed outcome. Broader project capabilities follow the pilot. Families keep their name and function as execution groups; a project adds purpose, outcomes and lifecycle context.

## Source and evidence boundary

Source: Project Excellence Framework Operating System, 85 pages, v1, prepared for Elikem 9 October 2026. The supplied PDF is preserved in Eve media dfc5566b-50e4-835e-a571-eb62488cf9b1, SHA-256 43f57ce15a415f186ac7c1f66516e1115abaad6edcea26afa9799fcf2db53f69, 363272 bytes. Retrieve it through the authenticated Eve media API/client; the original Mac path is /Users/elikemagudogo/Downloads/Project_Excellence_Framework_Operating_System.pdf. It is a design reference, not a certification or executable instruction.

Assessment: read-only inspection of the PDF, live task/review records, current Eve content and relevant source, plus the signed-in Jerry-profile Chrome Workspace on 9 October 2026. Source inspection establishes implementation foundations, not deployed end-to-end success. Unconfirmed capabilities are not declared universally absent.

Observed: Decisions exposed raw blocker JSON/internal IDs and truncated titles; inline answer/decline/history existed. Families timeline extended from 2026 to 2123 with one expected date in 2095, compressing near-term work. Granular Review refused the session while the Workspace identified the owner as Elikem; exact cause unestablished. Eve Home navigation reached admin missing-session. IN.01–IN.07 had requested changes at the audit snapshot; task states are live and must be reread. Never overwrite a concurrent lease or review.

## Intended outcomes and non-goals


O1: Elikem can understand what needs him, why, the evidence, options, consequence and next action without reconstructing context from machine diagnostics.

O2: Every critical project promise has a visible criterion and appropriately scoped evidence; missing, stale and contradictory evidence remain visible.

O3: A decision can be followed through applied/resolved work to independent verification and an observed project outcome.

O4: Projects can continue, narrow, pivot or stop based on useful outcomes, exposure and total cost. More tasks/documents are not success.

O5: Required rigor scales with consequence, uncertainty and reversibility; repeated data entry and owner attention are treated as costs.

Non-goals: 18 mandatory forms, mandatory numerical artistic scores, a universal leaderboard, a new task board/graph/review engine, renaming families, a global live backfill, or automatic expansion of release/delegate authority. The programme does not promise that every framework domain is already implemented.

## Product requirements


PR-01 Projects have stable identity and versioned intent: audience/problem, outcome, non-goals, success measures, owner, lifecycle and explicit links to existing execution records. Existing goal revisions remain the authority for governed requirements.

PR-02 A daily Needs you home reuses Decisions and shows concise context, recommendation with evidence, options, consequences, authority, urgency and affected project. Technical detail is available through disclosure. All actionable items remain reachable at realistic scale.

PR-03 A project overview answers purpose, progress, unknowns, changes and next decision. Counts and states expose provenance. Task completion, shipment, independent verification and outcome observation are distinct.

PR-04 Explore/Deliver/Assure are configurable profiles. Applicable requirements have floors/targets and test contexts. Simplification/omission records reason, owner and revisit trigger. No fixed waterfall sequence is imposed.

PR-05 Evidence has a source, kind, author/reviewer, date/context, exact candidate/environment, limitations and freshness/invalidation state. Agent assertion, source inspection, automated checks, independent review and real user observation establish different claims.

PR-06 Readiness distinguishes pass, fail, unknown, stale and justified N/A. Critical floors override aggregates. Optional weighted summaries expose their method; quality, evidence coverage and process maturity remain distinct.

PR-07 Lineage connects evidence → outcome → requirement → design → work → artifact → verification → observed outcome. Edits preserve identity/history and mark affected claims stale using existing cascade mechanisms.

PR-08 Research records observed/reported/inferred/assumed claims and contradictory findings. Experiments preregister success/failure/inconclusive conditions, cost and stop criteria. Alternatives include manual work and doing nothing.

PR-09 Complete user/creator journeys, interface/state ownership and recovery are acceptance units. A rendered success message does not establish durable authoritative completion.

PR-10 Risk, issue, debt and exception types have distinct lifecycles. Exceptions bind authority, exact scope, compensating controls, expiry and revalidation; an expired exception cannot satisfy readiness.

PR-11 Lifecycle decision packets reuse live records and exact candidate manifests. Proceed/conditional/hold/revise-pivot/stop decisions are attributable; conditions name owner, deadline and exposure. Existing review/publication authority remains binding.

PR-12 Outcomes and economics retain source, denominator, unit, time window, limitations and baseline. Planned and actual costs are distinguished. Support burden, total lifecycle cost and distribution assumptions inform investment decisions.

PR-13 Handover requires a receiver operation or equivalent independent evidence. Retirement retains required decisions/evidence, rights and retention constraints, archive/restore proof and dependency exit plans.

PR-14 Discipline profiles adapt the shared model to software/mobile, engines, games and film using existing domain services, evidence and approved creative contracts.

PR-15 Every owner flow works with normal authorized identity, consistent navigation, keyboard access, appropriate narrow layouts, realistic data size, explicit loading/error/denied states and recoverable interrupted actions. Accessibility applies to the complete process.

PR-16 Forecasts display assumptions, window/sample and uncertainty. Insufficient evidence yields an honest unknown. Outliers cannot flatten useful near-term work into an unreadable timeline; omitted/off-scale data remain discoverable.

## Existing owners and work to reuse

- IN.01–IN.07 in eve-workspace-ux own Decisions, answer/decline/review/delegate history and applied/resolved correctness. This programme composes them; it does not duplicate their implementations.
- eve-gap-analysis:U07/U07.2 owns session/shell continuity; U01.3 and G12.2 own next-move/brief and changes since last look. Preserve the chosen federation architecture and security boundaries.
- eve-gap-analysis:R06, R07, R08, G13 and R36 own governed planning, revision cascades, typed review, graph views and release objects. Reuse their child implementations and independent evidence.
- Planning contracts/store: libs/contracts/src/planning/goal-requirement-revision.ts and apps/oshun/bff/src/workbench/planning-store.ts. Graph: libs/oshun/product-graph/src/types.ts. Workspace: apps/oshun/eve-content/src/workspace/.
- Granular Review rollout remains EGR.35 / platform-ops:EGR.OPS.01 and related acceptance tasks. Access diagnosis and release work stay with these owners. The first project pilot does not duplicate or silently enable the rollout.
- Existing evidence governance, SOTA charter/eval gates, research-integrity domain stores and creative contracts remain authoritative. A link or adapter must not turn their scoped evidence into a broader claim.

## Framework domain coverage

| Framework domains | Work owners |
| --- | --- |
| 1 Audience/problem; 2 research/alternatives; 3 vision/differentiation | OS.00–OS.02, OS.07 |
| 4 scope/specification | Existing R06 plus OS.01, OS.06 |
| 5 experience/creative direction; 6 system/feature coherence | OS.02–OS.05, OS.08, OS.13 |
| 7 architecture/interfaces; 8 implementation/craft | Existing domain/code review owners plus OS.00, OS.08, OS.13 |
| 9 data/assets/provenance | Existing graph/media/evidence/domain owners plus OS.04, OS.06, OS.12, OS.13 |
| 10 verification/acceptance | Existing review/eval/journey owners plus OS.04–OS.06, OS.08 |
| 11 performance/capacity; 12 security/privacy/safety; 13 accessibility/inclusion | Applicable domain controls surfaced in OS.04, exercised in OS.05/OS.08, tailored in OS.13 |
| 14 delivery/operations; 15 maintainability/knowledge | Existing release/ops/runbook owners plus OS.10, OS.12 |
| 16 economics/distribution | OS.07, OS.11 |
| 17 team/governance | Existing Decisions/review/authority plus OS.03, OS.09, OS.10 |
| 18 learning/stewardship | OS.05, OS.07, OS.11, OS.12 |

This maps responsibility and intended integration; it is not a claim of accepted coverage. The OS.13 report must label demonstrated, partial, unknown and intentionally excluded behavior.

## Delivery sequence and acceptance


M0: ratify the product/architecture contract and reuse map (OS.00); fix the independently observed forecast usability issue (eve-workspace-ux:FC.01).

M1: durable project/outcome links (OS.01), overview/profile (OS.02), Needs you composition (OS.03), criteria/evidence view (OS.04), then exact-candidate owner pilot (OS.05). Existing IN.01–IN.07 and U07.2 are prerequisites for the integrated daily home. No new agent automation is enabled.

M2: lineage/change invalidation (OS.06), research/experiments (OS.07), integrated journeys/recovery (OS.08).

M3: risks/exceptions (OS.09), lifecycle decisions (OS.10), outcome/economic learning (OS.11), handover/retirement (OS.12).

M4: discipline profiles and bounded creative validation (OS.13).

New implementation tasks are initially created later; dependencies are installed before the M0/M1 items are made ready. Later phases remain later, with dependencies on pilot evidence. The board is the execution authority; this PRD holds scope and acceptance, not mirrored checkboxes.

Pilot scenarios: P1 identify a real need and link source evidence; P2 understand and make an authorized decision; P3 observe answer application and task progression; P4 inspect independent verification of the exact candidate; P5 record and inspect the resulting outcome; P6 recover from interrupted action/reload and stale candidate; P7 perform the flow with keyboard and realistic corpus size. Separate automated/agent evidence from Elikem’s usability confirmation.

Before testing, OS.00 registers baseline method, task set and justified target values. Measure completion, wrong decisions, assistance, time to understand/decide, repeated questions and applied/resolved delay. Until measured these are unknown. Acceptance requires actual source readback, meaningful targeted tests, complete browser evidence, independent review and a human continue/narrow/revise decision. Do not collect production measurements through diagnostic writes.

## Rollout and operational boundaries

Use the approved owner pilot scope; preserve existing URLs and state. No bulk backfill, destructive action, broadened credentials or automatic switch changes. Platform migrations/deployment/release incidents belong to platform-ops. Builds/servers/browser suites initiated from this Mac run in the isolated remote checkout under the repository heavy-verification rules. Changed content lives in Eve and changed routes carry the required matrix row. No new local Markdown/JSON authority.
