This guide is the curated front door to V8, Ariadne: The Self-Authoring Detective Universe. V8 connects the authored investigation world of V5 to the generative and agentic platform, but keeps one non-negotiable thesis: an LLM may propose the experienced surface; symbolic constraints and independent gates must prove the mystery fair and uniquely solvable before a player receives it.
V8 is an integration and generation-completion version. Its originating design documents describe the target; the derived backlog, actual symbolic core, compiler/runtime bridges, generated artifacts, tests, preregistered quality evidence, deployments, and release decisions establish completion.
Documentation map#
Drawn from its Mermaid source.
As text
Tree:
- V8 documentation
- Player promise
- Endless fair solvable cases
- Living interrogatable suspects
- Realized prose media and voice
- Continuity and open-world direction
- Solve-first pipeline
- Clew case proposal
- Minos constraint proof
- Palimpsest canon graph
- Anansesem and Loom realization
- Delivery
- Ori suspects
- Daedalus compiler and V5 bridge
- Theseus evaluation
- Oracle direction
- Truth
- Eight release-decision gates
- Gap and SOTA analysis
- Backlog dependencies and decisions
- Replay provenance and human evidenceShow source
mindmap
root((V8 documentation))
Player promise
Endless fair solvable cases
Living interrogatable suspects
Realized prose media and voice
Continuity and open-world direction
Solve-first pipeline
Clew case proposal
Minos constraint proof
Palimpsest canon graph
Anansesem and Loom realization
Delivery
Ori suspects
Daedalus compiler and V5 bridge
Theseus evaluation
Oracle direction
Truth
Eight release-decision gates
Gap and SOTA analysis
Backlog dependencies and decisions
Replay provenance and human evidence
Start with the question#
| Question | Canonical starting point | What it covers |
|---|---|---|
| What does the player receive? | V8 overview, feature hub, and feature index | Player experience, generated-case boundaries, living suspects, delivery, operations, and non-goals. |
| How is a case proven solvable? | Case/proof feature, symbolic-core architecture, and canonical contracts | Constraint proposal, uniqueness/completeness proof, fair-play constraints, canon, and typed verdicts. |
| How does a verified skeleton become a playable case? | Realization feature, surface realization, and compiler/eval | Prose/assets/voice, living suspects, compilation, V5 bridge, playtesting, and regeneration. |
| What directs delivery and operations? | Delivery/operations feature, cross-cutting feature, and Oracle/deployment | Direction, orchestration, gates, determinism, cost, localization, observability, rollout, and rollback. |
| What is real versus missing? | Gap analysis, product review, current code/tests/artifacts, and backlog | File-level reuse/new work and conservative completion state. |
| Which research and decisions govern the system? | SOTA research, orchestration decisions, and dependencies | External grounding, authoritative decisions, reused systems, new components, and providers. |
| Which gate vocabulary is canonical? | Architecture hub and Oracle/deployment | The eight ReleaseDecision gates and the separate seven-gate platform content harness. |
| Where are the visuals? | V8 diagram atlas and the global diagram gallery | Curated case, proof, data, lifecycle, ownership, generation, and release models. |
Case truth model#
This picture is wider than it can be shown here. Move across it, or read it as text.
Drawn from its Mermaid source.
As text
Parts: - Start (initial) - Proposed - ConstraintValid - UniquelySolved - FairPlayVerified - Realized - Compiled - Evaluated - Released - Rejected - Regenerate - Withheld - Retracted Connections: - Start to Proposed - Proposed to ConstraintValid: schema and world constraints pass - ConstraintValid to UniquelySolved: Minos proves one intended solution - UniquelySolved to FairPlayVerified: clue completeness and fair-play pass - FairPlayVerified to Realized: prose assets voice and suspects generated - Realized to Compiled: Daedalus emits versioned V5 case bundle - Compiled to Evaluated: Theseus and independent release evidence pass - Evaluated to Released: Oracle admits an eligible cohort - Proposed to Rejected: invalid contradictory or unsatisfiable - Realized to Regenerate: surface quality safety or canon failure - Regenerate to Realized - Evaluated to Withheld: required human-aligned or operational evidence absent - Released to Retracted: regression rights safety or canon conflict
Show source
stateDiagram-v2
[*] --> Proposed
Proposed --> ConstraintValid: schema and world constraints pass
ConstraintValid --> UniquelySolved: Minos proves one intended solution
UniquelySolved --> FairPlayVerified: clue completeness and fair-play pass
FairPlayVerified --> Realized: prose assets voice and suspects generated
Realized --> Compiled: Daedalus emits versioned V5 case bundle
Compiled --> Evaluated: Theseus and independent release evidence pass
Evaluated --> Released: Oracle admits an eligible cohort
Proposed --> Rejected: invalid contradictory or unsatisfiable
Realized --> Regenerate: surface quality safety or canon failure
Regenerate --> Realized
Evaluated --> Withheld: required human-aligned or operational evidence absent
Released --> Retracted: regression rights safety or canon conflict
Do not collapse proof, heuristic evaluation, independent human-aligned evidence, compilation, or live delivery. A generated solution is not a proof; solver SAT is not unique solvability; a fair skeleton is not a good experienced case; a Theseus diagnostic is not G5 evidence; a compiled fixture is not a released case; and a successful generation cannot bypass canon, safety, rights, cost, or rollback requirements.
Recommended reading journey#
Drawn from its Mermaid source.
As text
Steps: - 1. Resolve world canon cast motive and constraints (good). Propose. Clew, Palimpsest. Score 5 of 5. - 2. Emit typed replayable skeleton (good). Propose. Engineering. Score 5 of 5. - 3. Prove uniqueness completeness and fair play (good). Prove. Minos, QA. Score 5 of 5. - 4. Reject contradiction ambiguity or missing inference (good). Prove. QA. Score 5 of 5. - 5. Generate prose media voice and suspect behavior (good). Realize. Anansesem, Loom, Ori. Score 4 of 5. - 6. Compile and test the complete player path (good). Realize. Daedalus, Theseus. Score 5 of 5. - 7. Satisfy independent gates and stage cohort (good). Release. Governance, Release. Score 5 of 5. - 8. Observe retract roll back and update evidence (good). Release. Oracle, SRE. Score 5 of 5.
Show source
journey
title From V8 case idea to released mystery
section Propose
Resolve world canon cast motive and constraints: 5: Clew, Palimpsest
Emit typed replayable skeleton: 5: Engineering
section Prove
Prove uniqueness completeness and fair play: 5: Minos, QA
Reject contradiction ambiguity or missing inference: 5: QA
section Realize
Generate prose media voice and suspect behavior: 4: Anansesem, Loom, Ori
Compile and test the complete player path: 5: Daedalus, Theseus
section Release
Satisfy independent gates and stage cohort: 5: Governance, Release
Observe retract roll back and update evidence: 5: Oracle, SRE
Product, narrative, and puzzle design#
Start with the feature index. Separate immutable ground truth and deduction structure from regenerable prose/media/performance. Define what the player can observe, infer, challenge, and recover from; document how the system handles ambiguity, contradiction, canon conflict, generation delay, and unavailable modalities.
Symbolic, agent, and integration engineering#
Start with the architecture index. Trace typed case contracts through Clew, Minos, Palimpsest, Anansesem, Loom, Ori, Daedalus, Theseus, Oracle, and the V5 runtime bridge. Preserve seeds, versions, solver/proof artifacts, source/provenance, prompt/model/tool envelopes, and gate receipts so every release decision can be replayed and audited.
QA, governance, and operations#
Exercise unsatisfiable, multiply solvable, trivial, contradictory, misleading, unsafe, canon-breaking, rights-blocked, over-budget, localized, inaccessible, provider-failed, compiler-incompatible, and retracted cases. Validate both the symbolic skeleton and the full player experience. Current preregistered human quality evidence and production rollout posture—not a model judge alone—decide release readiness.
Page-set organization#
- Feature index — five consolidated topics covering promise, proof/canon, realization/suspects, delivery/operations, and cross-cutting pipeline/localization.
- Architecture index — six focused topics covering pipeline thesis, contracts, symbolic core, realization, compiler/eval, and Oracle/deployment.
- V8 diagram atlas — generation, proof, typed data, lifecycle, component ownership, player journey, and release.
- Root analysis/research/decision/dependency/backlog/review documents — why, evidence, binding decisions, delivery truth, and remaining work.
Documentation quality contract#
Every canonical page stays substantial, indexed exactly once, visually explicit, and precise about proof versus generation, immutable versus regenerable data, gate identities, evidence independence, versioning, provenance, and release state. Changes update contracts, solvers/compilers, tests/evals, backlog/decisions/evidence, and rollback procedures together. Generated integrity, freshness, search, gallery, and desktop/mobile checks pass.