Meaning Model MCP server

Read from the package source without running it. The installed server may list more.

• life_candidate_accept

  Atomically accept exactly one immutable complete candidate if its frozen parent is still the current head. This is consequential and should receive explicit human confirmation.

• life_candidate_annotate

  Append a plausibility judgment to the evaluation ledger without changing candidate or world canon.

• life_candidate_compare

  Compare two candidate endpoints and report whole-state divergence.

• life_candidate_observe

  Mechanically summarize only the Rust-projected fields retained on an uncommitted candidate, without writing a story.

• life_candidate_reject

  Reject one complete pending Rust candidate without changing the accepted world head.

• life_candidate_reroll

  Increment rollIndex and reroll all declared stochastic streams from exactly the same frozen parent and deterministic inputs.

• life_candidate_roll

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

• life_candidate_route

  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.

• life_engine_status

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

• life_estimation_proposal_inspect

  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.

• life_estimation_proposal_review

  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.

• life_estimation_request_createwrite action

  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.

• life_estimation_response_submitwrite action

  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 vali

• life_graph_query

  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.

• life_meaning_query

  Read a bounded page of authored concepts, abstract relations or cuts, referents, encapsulation cuts, events, event-referent bindings, physical cuts, and realization records from an optional Rust-stored Meaning Model layer. Filters are exact ids; this static administrative view of semantic data does not infer links, execute semantic records, apply cuts, or mutate the model, and must be protected by

• life_model_inspect

  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.

• life_model_register

  Validate, hash, and store one complete immutable revision-0 typed model in the Rust machine.

• life_model_revise

  Register a complete immutable successor model revision. Added dimensions and laws are schema changes, never in-place patches.

• life_model_validate

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

• life_modeling_context

  Return the paper-first reading order and minimum operational contract for story, person, observation, forecast, reconstruction, or counterfactual modeling. The live MCP process records access to both complete papers; content digests are provenance only and never replace reading.

• life_narrative_batch

  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. Read life-sim://protocol/narrative-understanding

• life_narrative_query

  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.

• life_narrative_register

  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 mu

• life_narrative_render

  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.

• life_narrative_revise

  Register one complete immutable successor of a Rust-owned narrative/understanding graph. Earlier text and testimony revisions remain addressable; no in-place overwrite occurs and the simulation world remains unchanged. The complete batch contract is documented at life-sim://protocol/narrative-understanding-graph.

• life_narrative_training_export

  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.

• life_profile_compile
• life_story_revision_diagnose

  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.

• life_trajectory_query

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

• life_trajectory_summarize

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

• life_view_query

  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.

• life_world_createwrite action

  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.

• life_world_inspect

  Inspect one canonical world head without changing it.

• life_world_refine_genesis

  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.

• life_world_revise

  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. T

• life_world_revision_inspect

  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.

• life_writer_contract_createwrite action

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

• life_writer_plan_evaluate

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

• Code scan16 source files scanned25/25
• –Live reliabilityno gateway calls yet and no remote to proben/a
• –Tool poisoningtools not inspected (local package is not executed); not countedn/a
• Auth qualitylocal package, no credentials required12/15
• Maintenancelast push 8 days ago15/15
• Maintainer identityregistry namespace matches repository owner7/10

Meaning Model: common questions

Is Meaning Model MCP server safe?

Yes, by our scan: it is graded A (91/100). Read the Meaning Model safety report

How do I install Meaning Model?

It runs on your machine. Copy the Claude Code, Claude Desktop or Cursor config from the install section.

Does Meaning Model need an API key?

No secret keys are declared. It reads 1 setting from the environment.

Is Meaning Model maintained?

The last commit was 9 days ago (2026-09-12). The latest release is v0.2.1.

What can I use instead of Meaning Model?

Servers from other publishers that do the same job: Strategic Agent Reasoning MCP server and Agent Memory Os MCP server.