Eve library

Canonical audit test coherence — generated

Evidence

Revision 1

On this page

Generated from the canonical verification record and test-coherence manifest. Tested source: b856ddd2fe3ff0f7ead8751554c21d7d8a010a20; 54 credited Vitest configurations. Proof classes are not interchangeable: no local test is promoted to real-provider, engine, or live evidence.

Summary beside pass counts#

Canonical check Tests passed Configurations sampled Proof classes Mutation/negative-backed
V1 @oshun/creative-autonomy 284/284 1 unit=1 1
V1 @oshun/agentic-studio 348/348 1 unit=1 1
V1 @oshun/studio-authoring 138/138 1 unit=1 1
V1 BFF autonomy bindings, E2E/route, and run lifecycle 66/66 1 e2e=1 0
V8 apps and case/canon packages 404/404 18 e2e=2, unit=8, integration=6, contract=2 15
V9 packages plus V9 canonical contracts 188/188 13 unit=6, integration=3, contract=3, e2e=1 9
V10 Rail apps, packages, and canonical contracts 782/785 (3 pending) 19 unit=18, contract=1 1

Per-configuration domain-value sample#

Configuration Recorded tests Class Domain-value claim/assertion Mutation/negative fixture Weakness Uncovered claims
libs/oshun/creative-autonomy/vitest.config.ts<br>libs/oshun/creative-autonomy/src/signals/signals.test.ts 284/284 unit Signal strength follows the configured half-life numerically.<br>Marker: toBeCloseTo(0.4, 10) never amplifies a future-dated observation Local component evidence; mocks and deterministic fixtures do not prove a deployed autonomous plane. real-provider execution; production placement; live user outcome
libs/oshun/agentic-studio/vitest.config.ts<br>libs/oshun/agentic-studio/src/budgets/budgets.test.ts 348/348 unit Budget rollover and consumption preserve exact period usage.<br>Marker: expect(consumed.usedInPeriod).toBe(100) fails closed for invalid budget requests and caps Local component evidence; mocks and deterministic fixtures do not prove a deployed autonomous plane. real-provider execution; production placement; live user outcome
libs/oshun/studio-authoring/vitest.config.ts<br>libs/oshun/studio-authoring/src/editorial-lifecycle/staleness.test.ts 138/138 unit Article age and source drift produce the expected staleness reasons.<br>Marker: toBe(DEFAULT_STALENESS_MAX_AGE_SECONDS.article) flags source drift even when the artifact is within its age window Local component evidence; mocks and deterministic fixtures do not prove a deployed autonomous plane. real-provider execution; production placement; live user outcome
apps/oshun/bff/vitest.config.ts<br>apps/oshun/bff/src/agentic/autonomy-bindings/autonomy-plane-e2e.test.ts 66/66 e2e A local autonomous cycle senses, produces, gates, and enqueues one review artifact.<br>Marker: submitted-review-window not sampled In-process E2E composes real local modules with injected providers; it is not a deployed-provider or browser/operator E2E. deployed provider execution; durable restart; human operator UI
apps/v8/daedalus-compiler/vitest.config.ts<br>apps/v8/daedalus-compiler/src/cli.test.ts 3/3 e2e Release compilation writes a manifest and distinguishes unasserted G4–G7 gates.<br>Marker: manifest.releaseGates.G4).toBe(false) never asserted Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
apps/v8/loom-service/vitest.config.ts<br>apps/v8/loom-service/src/adversarial.test.ts 29/29 unit Withholding an elimination clue prevents a unique-solution verdict.<br>Marker: viableCulprits.length).not.toBe(1) compiled case that lost a required deduction edge Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
apps/v8/minos-asp-sidecar/vitest.config.ts<br>apps/v8/minos-asp-sidecar/src/sidecar.test.ts 3/3 integration The HTTP solver returns the unique ground-truth culprit through clingo or DPLL.<br>Marker: expect(solved.matchesGroundTruth).toBe(true) returns 400 on a malformed body Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/v8/case-bundle/vitest.config.ts<br>libs/v8/case-bundle/src/case-bundle.spec.ts 7/7 integration A mixed generated/not-configured bundle records exact generated and absent counts.<br>Marker: provenance.generatedAssetCount).toBe(2) BLOCKS when the writer grades its own output Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/v8/case-csp/vitest.config.ts<br>libs/v8/case-csp/src/csp-solver.spec.ts 20/20 unit Pinned constraints yield the one exact culprit/weapon solution.<br>Marker: toEqual([{ culprit: 'ben', weapon: 'dagger' }]) returns no solutions for a contradiction Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/v8/case-gates/vitest.config.ts<br>libs/v8/case-gates/src/case-gates.spec.ts 7/7 contract Exactly seven required gates block unfair or underdetermined cases.<br>Marker: blockedGateIds).toContain('g1:fairness') no proven unique solution Contract/gate behavior proves admission semantics, not provider output, runtime placement, or production release. real provider artifact; runtime placement; production release
libs/yemaya/canon-graph/vitest.config.ts<br>libs/yemaya/canon-graph/src/consistency.test.ts 54/54 unit A wrong parent region becomes one hard geography contradiction.<br>Marker: out[0]?.severity).toBe('hard') wrong parent region Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-assets/vitest.config.ts<br>libs/yemaya/case-assets/src/budget-degrade.test.ts 64/64 unit Asset degradation never reserves beyond the configured ceiling and records drops.<br>Marker: reservedJobs).toBeLessThanOrEqual(ceiling) degrades under a tiny asset-jobs ceiling Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-compiler/vitest.config.ts<br>libs/yemaya/case-compiler/src/compile.test.ts 10/10 integration Compilation emits one valid evidence node per clue and binds generated image paths.<br>Marker: bound.Image).toBe('/Game/V8/Generated/doc1') not sampled Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-contracts/vitest.config.ts<br>libs/yemaya/case-contracts/src/contracts.test.ts 17/17 contract Difficulty bands and structural invariants reject out-of-band case specifications.<br>Marker: problems.some((p) => p.includes('targetSuspects')) not sampled Contract/gate behavior proves admission semantics, not provider output, runtime placement, or production release. real provider artifact; runtime placement; production release
libs/yemaya/case-director/vitest.config.ts<br>libs/yemaya/case-director/src/pacing.test.ts 33/33 unit Expert skill produces a harder case than novice skill.<br>Marker: expect(ex.difficulty).toBe('fiendish') a losing streak nudges it down Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-engine/vitest.config.ts<br>libs/yemaya/case-engine/src/clew.test.ts 22/22 unit Generated cases have exactly one viable ground-truth culprit across seeds.<br>Marker: expect(viable).toHaveLength(1) varies the culprit/structure across seeds Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-eval/vitest.config.ts<br>libs/yemaya/case-eval/src/eval.test.ts 14/14 integration Compiled cases solve to the intended culprit and a dropped deduction edge fails G4.<br>Marker: expect(solve.accusedId, `seed ${seed}`).toBe(culprit) G4 FAILS when the compiler drops a required deduction edge Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-localization/vitest.config.ts<br>libs/yemaya/case-localization/src/localization.test.ts 8/8 unit Localization preserves string IDs and fails loudly without a translator.<br>Marker: rejects.toThrow(/no translator/) catches a dropped string Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-pipeline/vitest.config.ts<br>libs/yemaya/case-pipeline/src/pipeline.test.ts 35/35 unit The shared budget clamps spend exactly at its ceiling without silent overspend.<br>Marker: spent('asset-jobs')).toBe(4) reserve never grants past the ceiling Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-suspects/vitest.config.ts<br>libs/yemaya/case-suspects/src/runtime.test.ts 34/34 integration The broker routes to offline without credentials and propagates live backend failures.<br>Marker: status.mode).toBe('offline') live backend that throws PROPAGATES Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-verifier/vitest.config.ts<br>libs/yemaya/case-verifier/src/integration.test.ts 26/26 integration The verifier proves one ground-truth culprit and rejects an ambiguous case.<br>Marker: matchesGroundTruth, `seed ${seed}`).toBe(true) catches an AMBIGUOUS case Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/yemaya/case-writers-room/vitest.config.ts<br>libs/yemaya/case-writers-room/src/writers-room.test.ts 18/18 e2e The local graph assembles scenes, sheets, interrogations, briefing, and banter end to end.<br>Marker: realized.interrogations.length).toBe(groundTruth.suspects.length) not sampled Deterministic or in-process case evidence; no current media-provider, Unreal placement, or player-quality result. real media-provider output; engine placement; live player outcome
libs/v9/aletheia/vitest.config.ts<br>libs/v9/aletheia/src/aletheia.spec.ts 19/19 unit The kernel computes the universe age in the 13.6–14.0 Gyr domain band.<br>Marker: c!.value).toBeGreaterThan(13.6) returns null for an unknown ref Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/v9/atlas/vitest.config.ts<br>libs/v9/atlas/src/grounded-leaves.spec.ts 25/25 unit A STEM fact binds one kernel reference and a 13.6–14.0 Gyr computed value.<br>Marker: ageGyr).toBeLessThan(14.0) rejects a STEM fact leaf with no kernelRef Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/v9/chiron/vitest.config.ts<br>libs/v9/chiron/src/chiron.spec.ts 12/12 unit Do-not-complete mode refuses graded full-answer requests.<br>Marker: expect(v.action).toBe('refuse') refuses to complete graded work in practice mode too Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/v9/cross-cutting/vitest.config.ts<br>libs/v9/cross-cutting/src/cross-cutting.spec.ts 17/17 unit Stage spend rejects a 20-token overspend above a 40-token cap after 30 spent.<br>Marker: ledger.spend('v9.stage.3', 20)).toBe(false) no silent overspend Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/v9/experience/vitest.config.ts<br>libs/v9/experience/src/experience.spec.ts 8/8 integration A wonder resolves to the expected concept while an unmatched wonder stays unresolved.<br>Marker: resolved?.conceptId).toBe('kalika:age-of-the-universe') no fabricated concept Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/v9/governance/vitest.config.ts<br>libs/v9/governance/src/governance.spec.ts 10/10 contract The registered V9 DAG has exact stages, dependencies, kill switches, and valid budget inheritance.<br>Marker: validateBudgetInheritance(reg)).toBe(true) not sampled Schema and release-gate semantics do not prove source entailment, provider quality, or learner outcomes. source entailment; provider quality; learner efficacy
libs/v9/hephaestus/vitest.config.ts<br>libs/v9/hephaestus/src/hephaestus.spec.ts 7/7 unit The Sun's computed right ascension sweeps 350–370 degrees over a year.<br>Marker: sunRaSweepDeg).toBeGreaterThan(350) does NOT reach the success state Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/v9/lesson-explorables/vitest.config.ts<br>libs/v9/lesson-explorables/src/lesson-assembly.spec.ts 22/22 integration Lesson assembly retains computed explorable, adaptive, retrieval, and 64-byte-hash provenance values.<br>Marker: provenance.explorableContentHash).toHaveLength(64) not sampled Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/v9/lesson-gates/vitest.config.ts<br>libs/v9/lesson-gates/src/lesson-gates.spec.ts 9/9 contract Seven required gates clear a valid lesson and block false or ungrounded claims.<br>Marker: blockedGateIds).toContain('g1:truth') asserting an unverified/false claim Schema and release-gate semantics do not prove source entailment, provider quality, or learner outcomes. source entailment; provider quality; learner efficacy
libs/v9/mnemosyne-glue/vitest.config.ts<br>libs/v9/mnemosyne-glue/src/mnemosyne-glue.spec.ts 11/11 unit FSRS stability is monotonic and matches the expected 0.4072 and 15.4722 anchors.<br>Marker: expect(easy).toBeCloseTo(15.4722, 3) a constant stub fails this Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/v9/prometheus/vitest.config.ts<br>libs/v9/prometheus/src/pipeline.spec.ts 8/8 e2e The local pipeline delivers one gated lesson with a 13.6–14.0 Gyr value and signed provenance.<br>Marker: ageValue!.value).toBeLessThan(14.0) not sampled Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/v9/theia/vitest.config.ts<br>libs/v9/theia/src/theia.spec.ts 20/20 integration The experience arc orders awe, understanding, and anticipation and ships a reachable widget.<br>Marker: arc[2]!.emotion).toBe('anticipation') not sampled Synthetic/local lesson evidence; no real-provider grounding, learner cohort, or efficacy result. real-provider grounding; live learner behavior; measured efficacy
libs/contracts/vitest.config.ts<br>libs/contracts/src/v9/__tests__/v9-contracts.spec.ts 20/20 contract V9 fact-node contracts reject absent grounding pins and classify STEM disciplines.<br>Marker: expect(v9NodeRequiresKernel('astronomy', 'fact')).toBe(true) REJECTS a fact node with zero grounding pins Schema and release-gate semantics do not prove source entailment, provider quality, or learner outcomes. source entailment; provider quality; learner efficacy
apps/v10/case-files-cook/vitest.config.ts<br>apps/v10/case-files-cook/src/cook.test.ts 10/10 unit The deterministic cook binds its edition to the requested local date.<br>Marker: expect(artifact.spec.caseId).toBe('case-files-2026-07-21') not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
apps/v10/rail-cli/vitest.config.ts<br>apps/v10/rail-cli/src/cli.test.ts 5/5 unit The dogfood CLI resolves the exact Linux Tauri app-data journal path.<br>Marker: '/home/rail/.local/share/app.oshun.v10.rail/dogfood-journal.json' not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
apps/v10/web/vitest.config.ts<br>apps/v10/web/src/components/MorningThread.spec.tsx 178/178 unit Rendering the morning Thread does not accept audio before an explicit press.<br>Marker: expect(onAcceptAudio).not.toHaveBeenCalled() not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-aphrodite/vitest.config.ts<br>libs/v10/rail-channel-aphrodite/src/live-channel.spec.ts 10/10 unit The adult-ring live channel emits a schema-valid manifest at its local boundary.<br>Marker: expect(ChannelManifestSchema.parse(result.manifest)).toEqual(result.manifest) not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-arete/vitest.config.ts<br>libs/v10/rail-channel-arete/src/channel.spec.ts 12/12 unit Arete's manifest makes absence non-punishing.<br>Marker: absencePolicy: { punishesAbsence: false not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-case-files/vitest.config.ts<br>libs/v10/rail-channel-case-files/src/case-files-channel.spec.ts 19/19 unit Pinning a canonical suspect produces the expected durable revision.<br>Marker: revision: 8, not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-motion/vitest.config.ts<br>libs/v10/rail-channel-motion/src/adaptive-program.spec.ts 34/34 unit Listener energy changes the selected deep-work program to the low-energy track.<br>Marker: ).toBe('focus-low') not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-nyx/vitest.config.ts<br>libs/v10/rail-channel-nyx/src/channel.spec.ts 11/11 unit Nyx rejects content produced after the requested observation instant.<br>Marker: cannot be produced after not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-ori-presence/vitest.config.ts<br>libs/v10/rail-channel-ori-presence/src/narrator.spec.ts 10/10 unit Ori reports the exact bounded call, token, and cost usage for a recap.<br>Marker: expect(result.usage).toEqual({ llmCalls: 2, totalTokens: 240, costUsd: 0.004 }) not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-stage/vitest.config.ts<br>libs/v10/rail-channel-stage/src/programming.spec.ts 79/82 unit Only the player-face compiler carries the favorite preference signal.<br>Marker: preferenceSignal: 'favorite', not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-tara/vitest.config.ts<br>libs/v10/rail-channel-tara/src/channel.spec.ts 12/12 unit Tara exposes exactly one bounded micro-act.<br>Marker: expect(manifest.microActs).toHaveLength(1) not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-v4-match/vitest.config.ts<br>libs/v10/rail-channel-v4-match/src/match-channel.spec.ts 10/10 unit A participant opt-out holds match commentary admission.<br>Marker: reasons: ['participant-opted-out'], not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-veritas-live/vitest.config.ts<br>libs/v10/rail-channel-veritas-live/src/channel.spec.ts 98/98 unit Veritas limits channel drips to claim-verified events.<br>Marker: expect(manifest.dripEventTypes).toEqual(['claim.verified']) not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-channel-wonder-recall/vitest.config.ts<br>libs/v10/rail-channel-wonder-recall/src/wonder-recall-channel.spec.ts 27/27 unit The review-due-wonders micro-act remains bounded to thirty seconds.<br>Marker: expectedDurationSeconds: 30, not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-kernel/vitest.config.ts<br>libs/v10/rail-kernel/src/daypart-programming.spec.ts 172/172 unit Daypart composition preserves its exact declared slot order.<br>Marker: expect(program.slots.map(({ id }) => id)).toEqual(['open', 'act']) not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-phase-a/vitest.config.ts<br>libs/v10/rail-phase-a/src/thread-composition.spec.ts 33/33 unit The Phase A invariant locks the Thread to the canonical six-slot order.<br>Marker: expect(program.slots.map(({ id }) => id)).toEqual(THREAD_SLOT_ORDER) not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/rail-thread/vitest.config.ts<br>libs/v10/rail-thread/src/thread.spec.ts 5/5 unit A fully sourced morning composes six ready slots.<br>Marker: expect(program.slots.map(({ status }) => status)).toEqual(Array(6).fill('ready')) not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/v10/test-channel/vitest.config.ts<br>libs/v10/test-channel/src/clock-quote-channel.spec.ts 4/4 unit The deterministic clock channel selects only a canonical catalog line.<br>Marker: expect(CLOCK_QUOTE_CATALOG).toContain(first.stateLine) not sampled Local policy, composition, and rendering evidence does not prove deployed source freshness, native OS behavior, or sustained human usefulness. deployed canonical programming; native release behavior; seven-day human outcome
libs/contracts/vitest.config.ts<br>libs/contracts/src/v10/channel-contract.spec.ts 53/53 contract The canonical contract rejects a micro-act longer than thirty seconds.<br>Marker: expect(result.success).toBe(false) expectedDurationSeconds: 35 V10 schema rejection proves admission semantics, not live media, production authority, or a human release decision. production source authority; live media delivery; human exit approval

Proof-class boundary#

  • Credited distribution: unit=35, e2e=4, integration=9, contract=6.
  • real-provider=0, engine=0, and live=0. Those classes require explicit external evidence references and cannot be inferred from local Vitest names or pass counts.
  • A configuration sample proves only its registered assertion. Every weakness and uncovered claim above remains visible beside the canonical count.