Meaning Model

Roll one complete, noncanonical successor from the current frozen head. It does not commit.

Rank scope-checked scalar projections of two or more pending candidates that share one frozen parent, interval, and dynamics. The result is advisory: it neither selects an in-world action nor accepts canon.

Record one conceptual decomposition as one step: the claim frame (what the concept is, what the carve is for, the range it claims), the climb to the root (each broader kind, what sets the kind below apart, its other kinds), every candidate part with its necessity, independence and universality verdicts and reasons, the completeness of the set (best checked against an outside list), real cases routed through it, and how the parts make the whole work. A candidate passing all three tests is a part; one failing only universality is a kind under a part (kindOf), not a part; one uncertain stays a fr…

Record which causes mattered most to an outcome, and how much. A question asks why an outcome Event came about rather than its contrast, the alternative it is set against (a different contrast picks out different causes), and reads the outcome on one measure, "probability" or an amount { unit, processId }, as it actually was. Each cause is an Event linked to the outcome (causes, enables, prevents or constrains; the kind follows the role when not given; or name an existing relationId), read on the same measure without it, everything else as it was, with its reason, an optional band and an optio…

Export a model-bound graph lineage as portable history: every bound model revision, explicitly declared external model_reference targets and author_reader life models with all their ancestors, the first graph revision in full, and later graph changes, with a bundle hash. Missing declared dependencies refuse export with repair guidance. Prefer destinationPath for a local JSON file: explicit absolute path, existing parent directory, no overwrite, maximum 480 MiB, compact receipt only. Without a path, inline output is capped at 1 MiB. Import rebuilds the same hashes. Needs accessScopes revealing …

Rebuild exported construction history on this engine. Supply exactly one of inline history or an explicit absolute sourcePath to a regular local JSON file of at most 480 MiB. File imports check the bundle checksum before any mutation. Register model and graph revisions in order, verifying each rebuilt hash, then replay with life_construction_replay. Source files are never modified; importing is a sequence of immutable registrations, not one all-or-nothing database transaction.

Replay how a graph and its model reached their present state, one graph revision at a time from the first: each step's reason, the model revisions it adopted and what they changed, the notes, reviews and prose it added, and each note beside the records it concerned as they were at that step. Levels: outline (one line per step), reasoning (full notes and model changes) or full (with prose and data). Page with offset and limit; focus on records or nodes to see only the steps that touched them. A modelHash replays the model revision chain alone. Read this first when continuing someone else's work…

Call this when your context was compacted or summarized, or whenever you are unsure what you still hold of what this server sent you. Results point to fixed parts they sent before (the standing questions, the scope note and the guidance) instead of repeating them; this returns the most recent copy of each now, with the models it came with, counts it as sent now, and lists what to read again to see the work whole. It changes no model or graph.

Draw one answer from a registered model's normalized Cut, typically a direction Cut over mutually exclusive continuations, with a recorded seed. The server computes u as the first 32 bits of SHA-256(seed) divided by 2^32 and takes the answer whose cumulative interval contains u, in the registered model's order: the engine stores a Cut's answers sorted by key, so the intervals run alphabetically, not in the order you wrote them (the result lists every interval). With record, the draw is stored in a graph bound to that model, and any earlier draw over the same Cut is reported and linked, so a se…

Load a whole UTF-8 book or document and automatically split it into ordered, typed source nodes in one atomic graph write. Supply text or an explicit local sourcePath, a documentId and title, and either modelHash or graphHash plus parentNodeId. Paragraph grouping and long-passage splitting are deterministic and preserve every byte; no AI or estimator is used. The document.source root and document.segment nodes carry source hashes and byte positions, keep access scopes, and are excluded from manuscript rendering and training. Imported words are source material, not accepted facts or world Event…

Project one exact document into reading order and UTF-8 byte positions, independently of world time. Resolve optional document.span metadata records anchored to stable node boundaries; positions follow text edits and retained split containers. Missing or reversed boundaries remain unresolved. Existing semantic connections stay open; no process values or reader responses are invented. Available without the storytelling add-on. Read life-sim://protocol/narrative-understanding-graph for the optional span record contract.

Describe the one authoritative Rust machine, its MCP interface, limits, and separately scoped coverage claims.

Description is built at runtime.

Inspect an immutable uncommitted estimation proposal. For proposals that include semantic model changes, includeProposedModel returns the complete administrative model definition needed for an explicit later life_model_revise call; data-only proposals contain no model. Protect model output with external authorization.

Record an immutable explicit review. Approval is refused if the bound world head is stale. Proposals containing a model are revalidated in Rust and return a separate life_model_revise step; data-only proposals have no model-registration step. Approved observed outputs return an exact forward ModelTransitionSpec observation fragment. Review itself never rolls, commits, registers, or mutates.

Create an immutable provider-neutral estimation request bound to one exact Rust model revision and accepted world head. The request exposes only the requested Rust-enforced projection and distinguishes assimilate, estimate, predict, infer, counterfactual, or simulate from reality versus creative intent.

Validate a provider response that disposes every requested coordinate exactly once as known, unknown, or unmodeled. Known coordinates require typed provisional Rust claims with uncertainty, provenance, authority, and evidence cutoff. A data-only response may omit proposedModel and proposalReason. Any semantic change requires a complete proposed successor ModelDefinition and reason, which Rust validates. This tool stores only an uncommitted proposal and never rolls, commits, mutates a world, or registers a revision.

Set the branches of a stage side by side: probabilities and unnamed mass, trajectories after the fork per process, each person's state per branch, Events each branch added, processes followed in some branches and not others, and siblings left thin. Computes no total. Read-only.

Fork one stage of sibling branches from one trunk revision and information cutoff: the answers of a forecast Cut (each premise spanning the forecast window, the probabilities carried where the Cut declares them for that window and cutoff and otherwise withheld with the reason, the remainder never branched), the options of a decision (with a waiting or no-action option unless noBaseline says why not; without either, the result asks), one or more assumptions, or a change to the past. Each branch is a model revision whose premise Event, narrowings and assumed realization carry a premise mark, wit…

List the stages and branches visible with these access scopes: premise, probability as issued (and joint probability inside a branch), baseline, branch graph and bound model, resolution, and whether the trunk has moved on since the fork. Read-only.

A bounded digest of one branch for switching between branches: premise, narrowings, probability, what the branch added to the trunk, values after the premise, people's states at a time, and the open consequence questions, each closed by modeling it or by a "sufficient here" note in the branch graph. Read-only.

Read the canonical Rust causal factor graph as a full immutable snapshot, a compact whole-graph skeleton, or a bounded neighborhood with every crossing edge and endpoint preserved. Neighborhoods are views and never replace or mutate the complete graph.

Record how much Events matter, and to whom. A scale names an audience, such as everyone who lives in this world, or those of them who follow one category such as football (concept names its Concept), and its levels from the most important down, each with an anchor that says what an Event must be to stand there; the go-to preset rarity has top_10, top_100, top_1000 and noted, counted over the whole span for that audience. Each judgment places one Event at one level of one scale, with a tag and an optional reason; judging it again replaces the level for this holder, and the revision history keep…

Define a lens: a way of seeing asked of many records, such as whether an act comes from fear or from love, or whether a person is wanting, bartering or standing for something. There are as many lenses as ways of understanding people and worlds; look for them in thinkers, traditions and disciplines, in the world being modeled, and invent the ones it needs. The lens is recorded as an Understanding Node of kind lens, linked to who or what it looks at. Its answers are Cuts on each record's Event with ids that begin lens.<id>., so they are weighted, keep a remainder, and can be estimated and drawn.…

Place selected lens readings beneath whoever holds them. By default use graph-defined lenses and built-ins already evidenced by historical readings; an unused built-in template creates no Events unless explicitly selected with lensIds. lensIds selects for this call only; repeat it or persist the intended lens with life_lens_define, without inventing numerical answers. An interpretation is not a fact of the world, so a reading does not sit on the record it reads: this adds the holder's understanding root (a context root of kind understanding) and one reading Event per record, contained by that …

Ask the active lenses of the model: graph-defined lenses and built-ins already evidenced by historical readings. In the storytelling profile fear or love is an optional template, returned in availableTemplates when inactive; select it with lensIds only if useful. This selection lasts for this call; repeat lensIds or persist the intended lens with life_lens_define. No numerical answers are needed to define a lens, and an unused template creates no open backlog. For each active lens, return applicable records without answers, their questions and Cut ids, changes in readings over a life, and ques…

Read again every stale lens reading of the records named, or of all records, across lenses, in one call: a reading is stale when its record was rewritten since it was read, or when it answered an earlier version of its lens. Each is asked the same question with the same answers and unit through the configured estimator, the lot applied in one model revision, and each reading's movement is reported, so a rewrite that changed nothing is seen to have changed nothing.

Read a bounded page of authored concepts, abstract relations or cuts, referents, encapsulation cuts, events, event-referent bindings, physical cuts, realization records, importance scales and event importance judgments from an optional Rust-stored Meaning Model layer. Filters are exact ids; text, which keeps records whose own words contain every word given; and from/to, which keeps records whose interval (half-open, so one that ends at from is left out) or moment falls in that span. fields "summary" returns one line per record (its name, time, region and the start of what it says), for scannin…

Read scoped attributed memory at an exact graph revision or the context's sole head. Never choose among branches. Filter by subject, holder, topic or literal text, with bounded pagination; includeSuperseded reveals retained earlier accounts. Returns process/Event/node links, source, uncertainty, time and holder judgments without treating them as verified facts.

Append attributed Understanding entries linked to native processes, Events or visible Understanding nodes in the chosen memory context. Use existing thought kinds for hypotheses, expectations, surprises, tensions, decisions and tests, not just facts. Preserve the subject, actual holder and source, uncertainty, and observation/validity dates separately from the authoring step. Stable entry IDs reconcile identical retries across restart. Explicit supersedes replaces an earlier account; a real change over time or disagreement need not supersede it. Advance only to an unambiguous descendant head, …

Description is built at runtime.

Capture explicitly supplied visible user and assistant messages as linkable source-document nodes when this context has opted in. Preserve exact text, supplied sequence, speaker and known timestamps. Never send hidden reasoning, system prompts or tool traffic. Stable IDs reject changed originals and reconcile exact retries; source messages are excluded from rendering, training and derived-memory query. Link interpretations with life_memory_record about or existing semantic links; source statements are not automatically true.

Explicitly enable or disable visible-message capture in this exact memory scope. Default off; disabling stops future capture and retains existing source messages. The caller supplies messages; the tool has no automatic access to its host chat. Settings are append-only history, separate from world state.

Read a paginated, scoped transcript at an exact graph revision or sole context head, with linkable message node IDs and visible incoming/outgoing edges. Supplied sequence is conversation order; unknown timestamps remain unknown. This reads stored source documents, never host messages or inferred current truth.

Write a snapshot of one stored model revision, its complete definition, to a new local JSON file, for a model larger than one result holds (clients cap a message at about 10 MiB). destinationPath is an absolute path whose directory exists; a file is never overwritten. Before writing, it checks that the file's text validates to this model's hash, and writes nothing if not. Returns the model's counts and the file with its bytes and sha256. life_model_register reads the file of a revision-0 model back with sourcePath; it refuses a later revision. A model's history, with its graph, moves through l…

Description is built at runtime.

Inspect a Rust-stored typed model summary. includeDefinition is an administrative operation: it returns the bounded complete definition, including process initial values, and must be protected by authentication/authorization outside this local MCP service. A result too large for one message comes back without the definition (its size and where to read it), and a summary too large on its own lists the first ids of each list, with listsCut counting them all.

Read the present state as a readable outline: description coverage (which Events carry numbers without a description), Things, the Event tree with descriptions and the Cuts under each Event, processes, concepts, understanding roots and documents. With a graphHash it overlays the Understanding Nodes linked to each record, at the chosen depth: none, count, first_line or full. Shown notes mark different or unknown model bases, with parsed writtenAgainstModel/currentModel in noteModelBases; these marks do not establish ancestry or invalidity. Use life_revision_check with exact models to identify a…

Read a model's own open questions, in any mode: what its structure shows is missing, inconsistent in time, undecided or unexplained, each as a question to answer by adding structure, with the tool to use. They go down the ladder (open what is coarse, give people whole lives, follow shocks into adaptations) and up it (the longer developments behind a moment, the concepts and laws Events instantiate). Also returns the model's jumps, where it changes most (the largest shifts, shocks, closest decisions and divergent readings), which is where a story or an explanation should look; and, with at, the…

Validate, hash, and store one complete immutable revision-0 typed model in the Rust machine. The result's descriptionCoverage lists Events that carry a Cut without a description; give each one, so its numbers mean something, and describe most other Events. requireDescribedNumbers refuses such a model. A complete valid request, model and graph are shown in life-sim://example/minimal-model-and-graph.

Register an immutable successor model revision, from either its complete definition (model) or its change from the predecessor (change): {reason, provenance?, upsert: {collection: [records]}, remove: {collection: [ids]}}, where collections are processes, laws, initial_claims, value_series, decomposition, dependencies and the Meaning Model collections (concepts, abstract_relations, abstract_cuts, referents, encapsulation_cuts, events, event_relations, event_referent_bindings, physical_cuts, realizations, normalized_cuts, context_roots keyed by event_id, temporal_cut_recompositions keyed by pare…

Validate a complete typed Life Simulation model in Rust without storing it.

Open a read-only local browser viewer. Open it early in mode live, with graphHash or modelHash, so the user watches the model fill as you build: the link follows saved graph revisions and their prose, and every model revision this server records after the one shown (each life_series_record among them), without reopening. Live follows only an unambiguous graph lineage, pauses at graph forks or insufficient scopes, and preserves reading context through guarded page refreshes; this is saved-revision following, not token streaming. mode snapshot (default) preserves the exact modelHash or graphHash…

Choose what the viewer shows a reader when it opens, and say why. Most of what a model records is not what a reader should see first: choose the rows that show how things change, the links and notes that explain it, and the moment and scale to look at. Curves alone hide why they move: keep the causal links (the ones you highlight are drawn bright, the other links of the rows on show faintly), your own notes beside what they are about, and a story's pages over the moments they tell (show prose); narrow the layers only for a reason. Give settings in the viewer's address vocabulary (show, rows, t…

Opens with two questions to ask the user before you build (askTheUserFirst). Return the reading order and operational contract for every mode: start-here, process breakdown, complete grammar, protocol, guide or profile, and an example, with full research papers as optional references for fuller explanations. Every fresh agent reads the grammar; use life_modeling_read when direct resource reads are unavailable. Workflow entries identify prompts separately from callable tools. Reading is instructed, not tracked or verified. Resource digests identify bytes, not comprehension.

Read canonical grammar or guidance through a tool, including in clients that do not expose MCP resources. Begin with life-sim://guide/start-here, which says what the Meaning Model is for and how to work in it, then read life-sim://protocol/grammar before first modeling, then the required resources listed by life_modeling_context. Follow nextOffset until null to read the complete resource, or read section "contents" and then the sections you need; pass expectedSha256 on later pages to detect changes. This tool does not attest comprehension or unlock mutations.

Description is built at runtime.

Atomically add one or many roots, nodes, and edges to an immutable Rust narrative/understanding graph. Rust creates a complete successor revision. On a nonempty graph every new-node component must connect in this same batch to an existing narrative node or validated stable anchor; a one-node batch is therefore valid when it includes that connection. The payload is {schema: life-sim-rust-narrative-batch/v1, previous_graph_hash, reason, provenance, add_roots, add_nodes, add_edges}; a complete, test-verified example is in life-sim://example/minimal-model-and-graph. Read life-sim://protocol/narrat…

Find drift between prose and the records about it: sentences that some earlier revision of the prose had and the current prose lacks, and the present-tense records (the bound model's Event descriptions, plans, disclosure records, writer's notes) that still quote them. Reviews, drafts, assessments, revision notes and superseded notes quote old text as history and are left out. A text match only: it finds copied or quoted sentences, not paraphrases. life_narrative_edit reports the same for the text it removes.

Atomically split, merge, move, reorder or edit existing narrative/understanding graph passages. Available for any modeled domain, without the storytelling add-on. Supply the exact predecessor graphHash, scopes covering its complete graph, a reason and ordered operations; the service preserves untouched records and constructs one immutable Rust-validated successor. Splits preserve exact rendered text and retain the original as a container; children can be subdivided again when finer independent attention is useful. Whole-part links can stay on the parent; linkAssignments can connect selected ch…

Confirm or correct a grounding proposal as the calling agent. Supply exact Event selections or a per-passage noLinkReason for every proposed passage. Atomically writes canonical grounding/renders Event edges or text-bound explicit reasons, preserving prose and model. Rejects stale graph heads, changed source fingerprints and incomplete decisions; never accepts suggestions automatically.

In one read-only call, propose Event links for existing unlinked rendered passages. Returns exact prose, native candidate Events, existing ancestor associations, explicit lexical-retrieval basis, omissions and source fingerprints. All unlinked passages are included by default, up to 200. No writes or external model keys; the calling agent must review and confirm or correct every selection.

Read an optional Rust-owned graph-native story/understanding artifact as a full graph, compact skeleton, or bounded neighborhood. Content and incident edges are removed when access scopes do not permit the node or edge; scopes are projection labels rather than authentication.

Rebind a model-bound narrative graph to a successor model revision as one complete immutable graph revision. Reads the whole graph with the supplied scopes, refuses partial projections and unrelated models, keeps every node including historical assessments, drops only edges anchored to predecessor model hashes, and submits the successor through the existing revise operation. Record fresh depth assessments against the new model before committing further prose.

Register a complete immutable revision-0 graph-native story/understanding artifact in the Rust authority. Story passages remain canonical nodes, externalized reflections remain distinct scope-guarded testimony, and typed edges may target stable model objects or validated nested subpaths. Read life-sim://protocol/narrative-understanding-graph before first use. This layer is optional and does not mutate the bound world. A complete valid request, model and graph are shown in life-sim://example/minimal-model-and-graph.

Render ordered story text from canonical Rust narrative nodes. The returned document is a projection with contributing node and content hashes; externalized reflections never render and no second story authority is created.

Register one immutable successor of a Rust-owned narrative/understanding graph, from either its complete definition (narrativeGraph) or its change from the predecessor (change): new or replaced nodes and edges, removed ids, and optionally new roots or a new source. Send the change for a large graph; the engine applies it to the stored predecessor, refuses it unless accessScopes reveal the whole predecessor, and validates the successor as a complete revision. Earlier text and testimony revisions remain addressable; no in-place overwrite occurs and the simulation world remains unchanged. The con…

Export deterministic text↔semantic-state records from a frozen Rust narrative graph and its exact model/world/candidate snapshot. These are single-snapshot alignments, not proof of cutoff-safe chronological order; causal training needs separately time-bound snapshots or a downstream mask. The tool performs no training. Externalized testimony is explicit content—not hidden chain-of-thought.

Evaluate bounded process questions with the configured Jev provider in one batch and automatically submit typed provisional claims to the core estimation exchange. Choice targets category/regime/distribution; Score maps a declared numeric rubric in the exact process unit, as the mean with its standard deviation or, with summary median, the median level with its interquartile levels; Noul targets an explicit probability process. Observations stay distinct from AI estimates. An answer that fails validation is declined on its own coordinate (disposed unknown, raw answer retained) and the rest of …

Record exact process-estimation values, evidence, questions and provider responses in a model-bound graph, together with an explicit core review and a real Understanding Node. Records a life_process_estimate proposal, or a data-only proposal submitted through life_estimation_response_submit, whose records keep each claim's holder, evidence type, cutoff and uncertainty. Each value becomes a node anchored to its process at its value time. To fill a model with dated values of its own processes, use life_values_record instead: it keeps the values in the model, not in the graph. Preserves the accep…

Compile scaffold profiles in Rust into an ordinary revision-0 ModelDefinition: person_scaffold for a whole life, change_arc_scaffold for a shock with its anticipation, focal change and adaptation, thing_scaffold for a machine, document or institution, relationship_scaffold, concept_scaffold, and the experimental Story, Person and Decision profiles. A scaffold is a starting structure to adapt, extend or replace with processes of your own; the modeler may also author a model directly with application-specific categories. Use life_modeling_context for the operational modeling guidance; the papers…

Record a review as an Understanding Node held by its actual reviewer (another model, a blind reader, an estimator, a person or a tool), with what the reviewer was given, how independent it was, the exact graph revision it read, the prompt, a hash of the rendered text it reviewed (checked against a supplied text hash), its verdict and findings, and links to what it concerns. Later changes that answer the review link to it with answers. When a review exposes an unmodeled assumption or an inadequate process, answer it by developing the model and bringing the dependent prose into line; otherwise r…

After a revision, inspect recorded dependencies that need review. Given the model before and after, it names detected changes (Events rewritten, retimed, relocated through region/substrate edits or removed; Cuts reweighted, withdrawn, removed or moved; processes whose declared state changed (initial value, bounds, unit, update mode or rubric); changed accounts of a process value (claims added, removed or changed, with holder and time), kept apart from its state; and Events whose numbers changed without their text: through a Cut on them, through a changed process state wherever they carry the p…

Discover saved graph-backed work in this engine without already knowing a hash. List immutable heads through the supplied accessScopes, including every visible alternative branch rather than choosing a latest version. Page through results, select an exact graphHash, then use life_construction_replay and life_model_outline before resuming. Includes visible root titles and replay/viewer calls; omits completely hidden work and private revision reasons. Models without a graph are outside this catalog. Check life_engine_status for persistence configuration.

Lay out the readings that would follow a subject's life at a step you choose: quarterly where the work needs detail, yearly or by life stage before it. Each slot nests in one existing reading of the same question, with the shares the slots inside it must average to, so you write what happened in each stretch and how the shares moved while the long view still holds. Use it when a question about a person or thing is read once across years in which things happen (open questions name those stretches, and life_series_record reports them as stillFlat). It records nothing; record the slots with life_…

Record a whole series of readings in one call: one question asked of one subject (a person, a company, a protocol, a market) at many dated intervals, each with its shares and its reason. Readings may nest, a long one with finer ones inside it, and must not partly overlap. Each reading becomes a dated Event inside the subject's life (or parentEventId), carrying its Cut, its tag as source, inferred, invented, exploring or sketch, and optional causes; the series is one model revision through the ordinary change path, with the remainder off unless series.remainder names a real unresolved share. Ch…

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Localize a cited story problem to the least foundational supported layer—model, cut, trajectory, or rendering—without rewriting prose, mutating canon, or pretending to measure literary quality.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Description is built at runtime.

Return a filtered and downsampled trajectory from the candidate projection already enforced by Rust. Large full paths remain outside normal tool output.

Compute a canonical Rust-owned piecewise-linear summary over a retained candidate subinterval. This is read-only and requires a full or decimated path.

Read named Understanding Nodes or reviews whole: each one's kind, holder, title, full text, data (a review's verdict and findings), provenance, parsed writtenAgainstModel and currentModel, and the links into and out of it (what it answers, what answered it). A different model basis leaves applicability unchecked; it does not establish ancestry or invalidity. A missing writing stamp is an unknown basis. Use it when the replay or the outline shows a note cut short; it reads up to 32 nodes without their neighborhoods.

Description is built at runtime.

Read dated values without reading the whole model: for each process, its values over a stretch of time with their bands and tags, as the model holds them, thinned evenly when long. With no processIds it lists the measures (ids, labels, units, how many values and over what span), filtered by text. Read-only; honors accessScopes.

Record the dated states of processes straight into the model: the working surface. A process is an empirical value with a unit or a defined scale with anchors (each point {time, value, lower?, upper?, tag, note?}), or a defined set of states (each point {time, state, tag, note?}; open it with states or a go-to preset: phase, intensity, direction, presence). One call takes many processes, at any times, past or present, and can open a new process under a parent as it goes (give process {label, unit, parent}), so populating a decomposition and deepening it are each one call; a process opened befo…

Query an explicit Rust-enforced projection of a canonical world or candidate. Empty requestedObservables returns metadata with no state; accessScopes are an access context, not authentication.

Create an isolated continuous world from a Rust-stored modelHash, or omit modelHash to use an explicitly selected or default approved preset. presetId and modelHash are mutually exclusive.

Inspect one canonical world head without changing it.

Build a general-purpose world model from a compact domain-defined scaffold without storytelling prerequisites. Automatically consider authored numerical judgments, useful conceptual openings and differences in meaning; the user need not request these separately. contextReview requires authoredJudgments, conceptualStructure and conceptVariation alongside broaderContext and longerTerm, with focalInterval, holder, represented/unknown/out_of_scope status and reasoned references. Missing context returns needs_context_review; missing considerations return needs_modeling_review before Jev calls or wr…

Replace an untouched Rust genesis world with an already registered direct-next model revision that preserves every existing process, edge, law, initial claim, and Meaning Model record exactly and may add authored detail. This is a genesis-only canonical mutation, not automatic discovery, cut execution, adaptive opening, or post-history migration.

Adopt a registered direct-next model revision at the current world time, including after accepted history. Requires the exact expected world hash. refine preserves existing records and state; revise explicitly permits compatible authored changes. Supply current values for every new process. Prior history and narrative sources remain frozen; old candidates cannot be accepted against the new head. This is an authored revision, not a simulated event or inferred migration.

Read a persisted authored world-revision receipt by its immutable hash. Frozen state is projected through requestedObservables and accessScopes; an empty request returns no state. This remains a trusted authoring session, not an authentication boundary.

Create a bounded writer constraint contract from one Rust-committed candidate. It cannot mutate canon.

Validate a structured writer disposition plan. Hard-field conflicts block rendering; renegotiation returns a model-revision-and-rerun request rather than changing canon.