Metis · Guides & deep dives

V9 Documentation Guide

Do not infer grounding from fluent citations, accuracy from model agreement, pedagogy from content length, completeness from a rendered widget, mastery from engagement, or release readiness from a generated lesson fixture.

6sections4 minread1table

On this page

This guide is the curated front door to V9, Metis: A Curious Ape's Guide to Reality—the consumer learning product that composes the Oshun knowledge, science, generative, agentic, simulation, and contemplative systems into a grounded journey from wonder to durable mastery. Its contract is not merely a beautiful answer: every claim is sourced or computed, every activity has a reachable success state, every lesson is pedagogically scoped, and every generated surface remains auditable.

V9 is an integration and productization version. Its originating product and feature documents set the target; the derived backlog, unified Atlas and lesson contracts, runtime composition, gates, tests, learner evidence, deployment, and operations establish what is supported.

Documentation map#

mindmap root((V9 documentation)) Learner experience Wonder resolution Chiron embodied teaching Hephaestus explorables Threads mastery and personal Atlas Lesson forge Atlas knowledge spine Prometheus solve-first generation Aletheia seven gates Lesson artifact and Theia continuation Composition Metis and Sophia grounding Mnemosyne learning science Nyx and Kalika computed truth Iris Yemaya Psyche Isis and Euterpe Delivery truth Product and gap analysis SOTA research and dependencies Backlog review and gate evidence Web-first operations and governance

Start with the question#

Question Canonical starting point What it covers
Why does V9 exist and who is it for? V9 overview, product analysis, and feature overview Learner, philosophy, product promise, experience pillars, business posture, and risks.
How does a free-text wonder become a scoped lesson? Atlas/wonder feature, Atlas architecture, and Prometheus feature Graph resolution, prerequisites, source pins, learner gaps, solve-first planning, and generation.
How is a lesson proven true and teachable? Subsystem/gates feature, seven gates, and lesson artifact Grounding, accuracy, pedagogy, completeness, quality, safety, provenance, and immutable evidence.
How does the learner interact and remember? Chiron/Hephaestus, explorable runtime, and mastery loop Socratic teaching, modality fallback, computed/generative explorables, success states, FSRS/IRT, and continuation.
What is new versus composed? Subsystem map/reuse and six-layer reuse stack Atlas/Hephaestus/experience additions plus the reused model, agent, tool, production, governance, and delivery layers.
What is actually complete? Gap analysis, product review, backlog, current code/tests/artifacts, and learner/release evidence Conservative status by subsystem, lesson stage, modality, cohort, and gate.
What evidence and dependencies govern decisions? SOTA research and dependency registry External evidence, shared systems, ownership, versions, new work, providers, and fallback posture.
Where are the visuals? V9 diagram atlas and the global diagram gallery Curated knowledge, lesson, mastery, data, state, composition, and release models.

Lesson truth model#

stateDiagram-v2 [*] --> Wonder Wonder --> Scoped: learner prerequisites and concept frontier resolved Scoped --> Grounded: sources and computed kernels pinned Grounded --> Planned: solve-first objectives activities and assessment exist Planned --> Realized: narrative media explorable and tutor surface generated Realized --> Gated: G1 through G7 produce versioned verdicts Gated --> Compiled: immutable verified skeleton and regenerable surface bundled Compiled --> Delivered: eligible learner cohort receives lesson Delivered --> MasteryObserved: assessment and retrieval evidence updates model Scoped --> Withheld: prerequisites sources rights or policy absent Realized --> Regenerate: quality modality or surface failure Gated --> Rejected: truth pedagogy completeness safety or provenance failure Delivered --> Retracted: source correction regression or policy change

Do not infer grounding from fluent citations, accuracy from model agreement, pedagogy from content length, completeness from a rendered widget, mastery from engagement, or release readiness from a generated lesson fixture. Numerical and scientific claims use canonical source pins or computed kernels; every explorable has a demonstrably reachable success state; gate verdicts are tied to immutable lesson versions; and learner-model updates preserve consent and auditability.

journey title From V9 wonder to durable mastery section Wonder Resolve intent prerequisites and learning goal: 5: Learner, Atlas Pin sources kernels and misconception risks: 5: Sophia, Kalika, Nyx section Forge Plan explain explore assess and continue: 5: Prometheus Realize tutor media and interactive surfaces: 4: Chiron, Hephaestus section Verify Run truth pedagogy completeness safety and provenance gates: 5: Aletheia, QA Compile versioned replayable lesson artifact: 5: Engineering section Learn Deliver accessibly and observe demonstrated understanding: 5: Learner Schedule retrieval and the next better question: 5: Mnemosyne, Theia

Product, curriculum, and experience design#

Start with the feature index. Define learner state, question, prerequisite frontier, objective, misconception risks, source/kernel authority, explorable success state, assessment evidence, modality fallbacks, and continuation. Preserve the science/human braid without presenting metaphor, generated narrative, or contemplative framing as empirical fact.

Knowledge, agent, and runtime engineering#

Start with the architecture index. Trace a wonder through Atlas, source/kernel grounding, Prometheus, Aletheia, media/explorable realization, Chiron, the lesson artifact, delivery, Mnemosyne mastery, and Theia continuation. Preserve concept/source/kernel/claim identifiers, seeds, model/prompt/tool envelopes, gate verdicts, artifact versions, and consented learner updates.

QA, pedagogy, accessibility, governance, and operations#

Exercise ambiguous wonders, missing prerequisites, contested or retracted sources, unit/numerical errors, impossible interactives, misconception traps, assessment leakage, unsafe/sensitive topics, age/region/entitlement denial, voice/avatar/WebGL/WASM fallback, localization, offline/partial behavior, provider loss, budget exhaustion, cache/version conflict, retraction, and rollback. Evaluate learning and truth separately from engagement and aesthetic quality.

Page-set organization#

  • Feature index — eight focused topics covering product overview, subsystem gates, wonder/forge, teacher/explorables, mastery, braid/commons/media, and governance/boundaries.
  • Architecture index — nine focused topics covering overview, reuse, Atlas, Prometheus, Hephaestus, lesson artifact, gates, and web-first cross-cutting delivery.
  • V9 diagram atlas — wonder/lesson flow, typed learning data, lesson state, composition, ownership, learner journey, and release.
  • Root product/gap/research/dependency/backlog/review documents — why, evidence, source boundaries, delivery truth, and remaining work.

Documentation quality contract#

Every canonical page remains substantial, indexed exactly once, visually explicit, and precise about sourced/computed/generated content, immutable versus regenerable artifacts, learner consent, gate versions, mastery evidence, and release state. Changes update contracts, knowledge/source/kernel data, tests/evals, backlog/research/evidence, retraction, and runbooks together. Generated integrity, freshness, search, gallery, and desktop/mobile checks pass.