MSGF V3.2 · Gated AI

Prefrontal cortex for AI
from IDE to cloud.

Six pillars, Grok-aware model presets, TRI majority CONVERGE when drift demands it, human-proof HITL with Session Replay and SIEM export, Sentry→Vault quarantine, Workspace SSO, and quantum-ready hybrid envelopes for vault secrets — plus a native Command Center that optimizes prompts locally and feeds a defensible token savings dashboard.

Try free 7-day Shadow Proxy Join beta Install the extension See what ships Explore full roadmap

Developer loop

Connect → optimize → verify → measure

01
Connect
Map project_origin · msgf_ide_* token · Command Center health goes green.
02
Optimize
0-token prompt with @-attachments and mandatory verify rules — local SOLO_FAST only.
03
Verify
Run Scripts or Safe Build · pass → Vault · repeat fail → Hall · deploy-gate.
04
Govern
Small Brain dual presets · TRI Big Brain on high drift · Shadow→Active gateway · P7 reputation prune · human-proof HITL · Session Replay · SIEM.

Network impact · marketing totals

AI that doesn't cost the Earth

Public network totals — not your private governance dashboard. We only publish environmental impact from proven avoided provider tokens — metered CONVERGE baselines when Small Brain skips cloud calls, or audited pack deltas from verify / Run Scripts. Estimates never inflate this board.

Live proven totals are building as tenants establish metered CONVERGE baselines. Until then we show zeros here — not illustrative placeholders — so marketing stays auditable.

Eco impact uses published coefficients applied only to proven avoided tokens.

Loading live metricsUpdated 1/1/1970, 12:00:00 AM

Opaque LLM stack

No gate before spend.

Prompts hit the model first. You pay for every call — no DEFEND check, no state prune, no lineage when something fails.

Prefrontal cortex for AI

DEFEND / GATE before model spend.

MSGF runs pillar checks, Vault/Hall memory, and consensus before tokens burn — so compliance and engineering share the same truth.

Analogy

Human brain vs MSGF

The same control loops biology uses — mapped onto gated AI infrastructure.

Human brainMSGF

Prefrontal cortexDEFEND / GATE
Working memoryHot SHARD / State Ledger
Long-term learningVault + Hall
Habit vs deliberationSmall Brain vs Big Brain
Error memorySource reputation (P7) + quarantine

Shadow mode

Prove projected savings in Shadow mode

Zero-latency pass-through to your provider. MSGF evaluates optimizations in the background and records projected savings — flip to Active when the ROI is credible.

Start a free trial — we mint your Shadow Proxy key. The 7-day clock starts on your first call (unused keys expire in 14 days).
Point OpenAI or Anthropic SDK baseURL at MSGF with x-msgf-mode: shadow.
Watch the proof ledger on your trial dashboard; get the email report when the window ends, then start 3-day Individual Pro full access.

OpenAI · baseURL …/api/v1

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY,
  baseURL: "https://elphiesgatedai.elphiesyntax.com/api/v1",
  defaultHeaders: {
    "x-msgf-mode": "shadow",
    "x-msgf-key": process.env.MSGF_LIVE_KEY!,
  },
});

Anthropic · baseURL …/api

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.ANTHROPIC_API_KEY,
  // SDK appends /v1/messages — use /api (not /api/v1) as baseURL.
  baseURL: "https://elphiesgatedai.elphiesyntax.com/api",
  defaultHeaders: {
    "x-msgf-mode": "shadow",
    "x-msgf-key": process.env.MSGF_LIVE_KEY!,
  },
});

Start free 7-day trial Token Savings (signed in)Full getting started

Shipped for production dev

Not a roadmap slide — these capabilities run in Cloud Run and the Pulse Guard extension today.

IDE Command Center

MSGF Pulse Guard sidebar: connection health, prompt optimizer, Run Scripts, Safe Build (async preflight), and advanced heal ops in Cursor or VS Code.

Vault · Hall verify loop

Allowlisted execFile runs sync pass to verify-result and fail to dev-event. Learning accrues without burning Big Brain on every build — then unlock deploy-gate.

Model presets + Grok

Pick your Small Brain pair: Claude+Gemini (default), Claude+Grok, or Gemini+Grok. Big Brain uses TRI majority (Claude + Gemini + Grok) when drift is high — humans notify only above the high-drift threshold.

Defensible token savings

Metered provider usage, proven avoidance (not estimates), weekly/monthly Reports + PDF, and Shadow Proxy projected $ kept separate from public eco.

Shadow Proxy → Active Governance

Point OpenAI or Anthropic SDKs at /api/v1. Shadow mode proves projected bill savings and would-have P7 promote/block counts with zero added latency; Active mode applies cache, state-gating, sharded prompts, and live reputation — license-bound auth, never spoofable tenant headers.

Sentry → Vault quarantine

Sentry crashes match Vault wins and mark them QUARANTINED (no silent auto-Hall) and write a P7 demote. Ops restore or demote on /admin/ops — runtime monitoring links to governance memory.

Workspace SSO

Company domains and Google Workspace SSO for corporate seats. Consumer Gmail lands on invite-only — not a silent tenant.

Platform & enterprise

TRI majority CONVERGE

Optional T1→T3 cost ladder for risk paths. Platform Big Brain: Claude + Gemini + Grok majority when enabled. Model splits alone no longer force HITL when majority agrees.

Signed HITL audits

Skip-MSGF and ARBITRATE decisions leave HMAC-signed, hash-chained snapshots — a compliance-ready trail of who approved what on /admin/ops.

Deploy gate

CI checks GET /api/msgf/deploy-gate for your project_origin. Ship only when verify is green for that silo.

Compound tenant isolation

Every vector scoped by project_origin + subpath — agencies and monorepos keep client and app memory from bleeding across silos.

Quantum-ready security

Optional hybrid post-quantum envelopes (X25519 + ML-KEM-768) for vault secrets and ML-DSA-65 authorship certificates. Frontend HTTPS PQ only when the load balancer SSL policy is ENABLED — not Pulse, Redis, or Supabase.

Admin ops control tower

One /admin/ops surface: audit hub timeline, Session Replay + harm ledger, most-used resources, model fitness, diff impact, tenant budgets, SIEM export, trusted-OSS bulk ARBITRATE, and Sentry quarantine.

Platform governance audit

Hashed resource-usage rankings, unified platform_audit_events, prompt-session forensics (purge-exempt), and human-proof gates — RED and harm always open HITL; budgets never bypass safety.

Budgets + SIEM export

Hard monthly dollar caps and rapid-retry circuit breakers before gateway dispatch, plus async OpenTelemetry JSON webhooks to customer SIEM — primary paths never wait on export.

Six pillars. One framework.

MSGF V3.2 — modular state gates that never mix contexts. Full pillar guide

Human Authorship Ledger

Rhythm telemetry and HAL signals — how input arrived, sealed for every gated interaction.

Flow & Consensus

Ordered gates: ingest, verify, dual or TRI-majority arbitrate, persist — no step-skipping.

Entity Profiles

Tenant roles, entitlements, and credit metering tied to sovereign identity.

State Ledger

Hot-layer beats and revision locks — nanosecond validation before AI acts.

Narrative Audit

Forensic bundles and audit lines you can read, not hidden model chatter.

Vault & Hall

Proven fixes in the Vault; failed logic in the Hall — lineage you can inspect.

Elphie Syntax product family

Other products on the same brain

MSGF is the prefrontal cortex. Author Ecosystem and Syntax Education are sovereign surfaces powered by the same gates — explore from elphiesyntax.com.

Sovereign · For writers & publishers

Foundational testing

Author Ecosystem

Sovereign narrative infrastructure — document ingest into wiki & outline, HAL proof-of-authorship, Vault Pact NDA, MSGF-gated Librarian & Critic, and Cool Down locks on the Phase 2 path. ~48% overall across three phases; foundational testing on authorecosystem.

Phase 1 ~78% — ingest (~90%), project switcher (100%), HAL cert, MSGF Pulse bridge
Phase 2 ~42% — Cool Down locks, bicameral audit, editor suite, Community Guild
Five tiers Free → $199.99 · Publisher Hub discovery on the Phase 3 horizon

Join foundational testing

K–12 · LTI 1.3 · Utah-aware

In development

Syntax Education

Socratic sandbox with grade-aware AI Allowance and district-approved curriculum slicing — on the shared MSGF brain, isolated by tenant. Code ~78% of full roadmap; host and public registration still in development.

Vite shell + lessons, board, curriculum, governance UI wired in repo
Classroom OAuth + Docs add-on need production env soak
Public host syntaxeducates.elphiesyntax.com — not open for signup yet

Public signup not open yet.

Full platform chooser

Ready to gate your AI?

Map a repo, install Pulse Guard, and open your governance dashboard with live token savings.

Try free 7-day Shadow Proxy Sign in to dashboard

Prefrontal cortex for AIfrom IDE to cloud.