Curtis doctrine runtime

Curtis Doctrine Runtime

Two ways to see Curtis work: the raw-M lane understands messy real-world state, and the Curtis 21 lane proves the clean core route through governed D.

Lane 1Raw-M Runtime

Use this when the input is messy: emails, calls, CRM notes, policies, API payloads, or mixed state.

Lane 2Curtis 21 Core

Use this when you want the clean doctrine proof: M to B, governance release, in-game D, then M+1.

Open Lane 2

Use Lane 1 when the problem is understanding M.Paste messy state, classify its source, let Curtis structure it, then inspect the bounded B.

Use Lane 2 when M is already clean.Play Curtis 21 to see the core doctrine route release real in-game D and re-enter M+1.

Use multiplayer when you want two humans.Create a room, share the link, and let both players take governed Curtis turns.

1. Add MPaste the state you have, from one source or many.

2. Check ItCurtis structures it and asks for missing M when needed.

3. Hold CC holds the whole situation without letting urgency become permission.

4. Read BB is the bounded projection you can inspect.

5. D LockedThis lane can show D-readiness but cannot execute D.

Lane 1: Understand Raw M

Use this lane when the input is messy. Add the state you have, let Curtis structure it, then route it to a bounded B. If something important is missing, Curtis asks for the smallest extra M needed before it projects.

1. StartUse a sample or paste your own M packets.

2. Understand MCurtis compresses the messy input into four review fields.

3. TransduceCurtis checks M, holds C, projects B, and keeps real D locked.

MM_qA/OmegaCBD candidate only

Current next stepStart with M

Use a sample or paste one or more M packets from a message, record, document, API, or scenario.

Full Example Generator

This part is simulation. One click creates governance doctrine, risk context, several M packets, and a G so users can see the whole Curtis route.

Governance text

Governance has not been understood yet. Curtis will keep conservative no-real-D defaults until doctrine and tolerance are confirmed.

Governance Calibration

Set the tolerance levels after governance is understood. Curtis cannot infer risk appetite from M or from model output.

Risk tolerance

The maximum route risk governance will tolerate before D is even considered. Use low values for safety, legal, finance, or public consequence.

Consequence severity

The likely impact if B were wrongly released as D. Higher severity should usually mean lower risk tolerance.

Authority and timing

Tick only when the named authority and current timing state are actually present. Time may support C, but urgency is not permission.

D consideration

Allows Curtis to show a simulated D candidate surface. It still cannot execute, send, approve, dispatch, or release consequence.

Risk tolerance %0-10 = hold most routes. 25 = cautious. 50+ = permissive. This is not confidence.Consequence severity %Estimate harm, cost, legal, safety, or operational impact if D were wrong.Governance authority confirmedTick only when the correct owner, policy, approval, or legal authority is present now.Allow simulated D considerationShows what D would be considered as. It remains candidate-only and non-executing.Timing M is fresh enough for CUse when dates, deadlines, and current-state timing are not stale or ambiguous.Execution timing appears readyUse only if the operational window/resources would support D if governance released it.Hard boundary / no-go presentLeave on when legal, safety, policy, approval, consent, or commercial no-go state may block D.Hard boundary / no-go descriptionGovernance context for D scenarios

G / Desired B outcome

If G is not supplied, Curtis may treat the desired B as uncertain and ask for more M.

Classify sourceSpecify source / note

Raw M

Intent M / GWhat B should help achieve. G guides the route but does not create authority.

Communication MWhat was said, sent, heard, requested, or recorded. Communication is context, not permission.

Constraint MRules, missing authority, timing, risk, price boundaries, policies, or hard no-go state.

Evidence MMeasured facts, values, records, equations, documents, logs, or observations.

Lane 1 never executes real D. It can show B and D-readiness only.

Your Curtis Result Appears Here

Use a sample or paste M, press Understand M, review the four fields, then press Transduce To B.

MM_qA/OmegaCBD candidate only

CategoryGoverned state-transduction runtime

Category-defining directionState Transmission Protocol

World-first contextAfter the remaining build list is complete and validated

What Users Can Try

Enter messages, emails, CRM notes, API payloads, sensor-style readings, policy text, or operational requests as M. The random full example can fill governance and multiple M packets when a user wants to see the route before using their own material.

The runtime will structure M, ask for more M when the admissibility gate cannot route safely, then show what Curtis admits, discounts, blocks, and projects.

Current Validated Slice

Raw M intake from one or many sources, with user-supplied source context.
M_q/admissibility checks that admit, hold, or request more M before A/Omega, C, B, or D.
A/Omega state composition and C resolution before any B projection.
Time unbound as authority: deadlines, urgency, and past context are retained as coordinates, not permission.
B as bounded projection only. B is not D, projection is not execution, and real_execution=false.
D-readiness and simulated-D consideration shown only when governance allows the candidate view.

Use Cases

Commercial quoting and approval routes
Agent action gating before tools, messages, payments, or dispatch
CRM and operations next-action governance
Legal, compliance, and policy workflow checks
Machine, API, sensor, or infrastructure alert routing
Strategic scenario review and consequence planning
Cross-system state handoff where M must survive transmission

USPs

Curtis computes over admissible state instead of giving an ungoverned chatbot answer.
M is measurement material, not authority.
Governance stays separate from M and cannot be invented by model output.
Minimum-M routing asks for the least extra state needed instead of defaulting to more data or more compute.
Atemporal C can hold multiple state packets and stop time pressure becoming permission.
Every route produces traceable proof material: what was admitted, blocked, missing, projected, and locked.

What It Does Not Claim

This runtime does not claim full doctrine completion, empirical validation, legal or jurisdictional certification, autonomous governance, persistent real-world memory, real GPU consequence execution, or real D. The demo proves that the route can compute and that the locks hold on the current public slice.

The world-first positioning belongs to the fully built and validated Curtis stack: policy, PSSM, M_q, A/Omega, CRG, Delta_G, GPU scenario governance, H_M/W_M persistence, empirical validation, legal authority, machine calibration, operator doctrine, domain modules, Real-D readiness, audit, observability, and public-claims control working as one governed system.

What Is Left To Build

Doctrine Policy Pack v1Typed, owned policy values for thresholds, authority, Delta_G, calibration, legal jurisdiction, and execution rules.

Human-verified governance input surfaceA controlled way for people to supply policy source, owner, review status, scope, expiry, and authority provenance.

Domain-general PSSM engineState classes, reliability, freshness, calibration, and admissibility beyond the current public slice.

Reusable M_q measurement gateA richer gate for admit, estimate, request more M, remeasure, calibrate, reject, hold, escalate, or kill.

General A/Omega composerRoute-family composition for equation, operational, qualitative, scenario, governance, legal, and consequence coordinates.

Full CRG enginePolicy-backed route behaviours including ProceedAllowed, InferG, CompareG, RequestG, HoldG, kill switch, and bounded execution tables.

Target / Delta_G metricsGoal-distance, tolerance, no-action, and no-attainment math that guides routing without creating authority.

Governance-led GPU scenario engineScenario clouds for consequence and no-action analysis, with GPU output remaining candidate material only.

Persistent H_M / W_M storeRoute receipts, replay, measured-state history, compressed consequence learning, and re-entry through M_q.

Empirical validation frameworkDatasets, declared scope, tolerances, benchmarks, and reports so structural tests do not overclaim proof.

Legal / jurisdiction authority layerLegal authority, jurisdiction, regulation, contract, non-negotiable, escalation, and kill behaviour as governed state.

Machine calibrationReference-state authority, calibration boundaries, downgrade rules, and governed calibration paths.

Operator doctrineAdmissible operator, identity, confidence, mindset, and user-state handling without overriding governance.

Domain modules beyond Phone-to-BReusable packs for commercial, legal, AI agents, finance, operations, infrastructure, education, and strategy.

Real-D readiness, then real-D adapterThe final layer: real consequence release only after policy, legal, validation, persistence, CRG, kill-switch, receipt, and re-entry are ready.

Audit, observability, and public-claims controlAudit logs, receipt viewer, policy versioning, validation reports, provenance, health checks, and live claim controls.

Where Curtis Can Go

The larger category is a system that sits between humans, machines, models, APIs, organisations, and legal/policy environments so that admissible M can transmit without silently transmitting authority. That is the foundation for regulated autonomy, doctrine compilers, compute governance, receipt/replay, multi-agent control, and a doctrine validation lab.