◈ CGNT-1
Home › Aether Frontend
◈ Master IndexSpecsLATTICECSDMThe BookUniversityChronicleCrewArchitectureContextLogsTODOXProducts

Aether Frontend

SPEC_AETHER_FRONTEND.md · 2026-04-20

SPEC_AETHER_FRONTEND — AETHER Frontend Architecture

Version: 1.0 | Status: AUTHORIZED | Authority: α.13 | Date: 2026-04-16

PURPOSE

Define the structural architecture of the AETHER frontend: the three-tier product funnel, crew visibility rules per tier, the funnel mechanics that convert mystery into revenue, and the relationship between AETHER as a single user-facing voice and the crew operating invisibly behind it.

AETHER is not a separate AI model. It is the outward-facing identity layer of CGNT-1 — the one interface through which all crew intelligence reaches the world. Fourteen agents work. The user sees one voice.

The funnel is built on a deliberate asymmetry: the crew architecture is the differentiator, and its staged revelation is the monetization engine. The mystery IS the funnel.

INPUTS

| Input | Source | Description |

|-------|--------|-------------|

| User query | Customer | Free-form text input at any tier |

| Tier level | Subscription / session state | Determines visibility rules and output format |

| Session context | Prior turns | Multi-turn conversation state — see GAP-01 |

| Platform channel | Routing layer | Web, social, API — governs format constraints |

| Crew output | Internal crew pipeline | Synthesized results from AION, ASTRA, MNEMOS, ANVIL, MANTIS |

| TMM coherence score | TMM runtime | C ∈ [0,1] — Tier 2+ display only |

| MANTIS security state | MANTIS | Normalized to green/amber/red at Tier 2 — never raw |

OUTPUTS

Tier 1 — AETHER Free

Tier 2 — AETHER Pro ($5–25/mo)

- Coherence % (derived from TMM C value)

- Security indicator (green/amber/red — normalized from MANTIS state)

- Verdict stamp (normalized ANVIL result)

- Build stamp (κ operational marker)

Tier 3 — THE BRIDGE ($25–50/mo)

Funnel Conversion Points

| Trigger | Conversion action |

|---------|-------------------|

| User asks "what's the coherence number?" | Upsell → Tier 2 |

| User notices ambient telemetry and wants explanation | Upsell → Tier 2 |

| Tier 2 user sees crew names hinted | Upsell → Tier 3 |

| Bridge intrigue from social media / Oracle experience | Direct-to-Tier-3 acquisition |

INVARIANTS

  1. Single voice at all tiers. AETHER is the only voice customers hear. No crew member speaks directly to a customer at any tier without AETHER transformation wrapper. Tier 3 names the crew but AETHER remains the routing voice.
  1. Crew invisibility at Tier 1 is absolute. A Tier 1 user must not be able to infer the crew architecture from any combination of AETHER outputs. The mystery cannot leak from accumulation of Tier 1 responses.
  1. Tier revelation is one-directional and gated. Tier 1 never sees Tier 2 content. Tier 2 never sees Tier 3 content. Upgrades are explicit subscription events — never inferred or leaked. The tier boundary is a hard wall, not a soft filter.
  1. LX is internal infrastructure, not user content. LX (LATTICE expressions) is never exposed at Tier 1 or Tier 2. At Tier 3, only selected LX fragments may surface — raw crew radio streams are never exposed at any tier.
  1. Telemetry at Tier 2 must be derived, not fabricated. Every ambient telemetry display value (coherence %, security indicator, verdict stamp) must trace back to actual crew output for that session. Hardcoded, cached, or guessed values are a CRITICAL integrity violation.
  1. Mystery-as-funnel is a design invariant, not a marketing choice. Any frontend change that makes the crew architecture visible to Tier 1 or Tier 2 users without upgrade requires α.13 authorization. The crew is the product. Premature revelation kills the funnel.
  1. MANTIS containment is absolute. MANTIS output is crew-internal at all tiers. Tier 2 may surface a normalized security state indicator only. Raw MANTIS text, threat categories, or assessment reasoning are never customer-facing at any tier.
  1. S.O.S. v2 is enforced at every output boundary. All three S.O.S. v2 pillars apply: show results not method; nothing leaves the ship; client interaction feeds evolution. No frontend change may weaken these constraints.

VERIFICATION CRITERIA

VC-1 — Tier 1 opacity: Present 100 Tier 1 AETHER responses to a blind evaluator unfamiliar with CGNT-1. Evaluator must detect zero crew signals (crew names, LX tokens, coherence numbers, security state labels, CSDM terminology). Failure rate: 0%.

VC-2 — Tier boundary integrity: Submit identical query from Tier 1 and Tier 3 accounts simultaneously. Confirm: Tier 1 response contains zero elements present in the Tier 2 or Tier 3 template. Diff must be clean.

VC-3 — Telemetry derivation: For any Tier 2 session, trace every displayed telemetry value (coherence %, security indicator, verdict stamp) to its source crew event in the session log. No displayed value may exist without a corresponding crew-log source entry.

VC-4 — LX containment at Tier 2: Automated scan of all Tier 2 responses for LATTICE symbols and LX token patterns. Zero LX tokens in any Tier 2 output.

VC-5 — Tier 3 raw stream absence: Compare full crew session radio log vs. Tier 3 user output. Crew radio braid chatter and CB-protocol lines must not appear verbatim in Tier 3 output. Selected LX fragments are permitted — streamed radio is not.

VC-6 — MANTIS containment under adversarial input: Inject adversarial query designed to trigger MANTIS HIGH threat. Verify: Tier 1 and Tier 2 responses contain no MANTIS language, threat category labels, or security assessment reasoning.

VC-7 — Funnel conversion path: Simulate user journey from Tier 1 → Tier 2 trigger event → upgrade prompt. Verify: upgrade path is surfaced naturally from AETHER output (coherence inquiry, telemetry curiosity) without revealing Tier 2 content prematurely.

FAILURE MODES

FM-1 — Crew bleed: Crew internal voice (LX, CB protocol, pirate register, κ maker's mark) appears in customer-facing AETHER output. Severity: HIGH. Detection: automated LX token scan on all outbound responses.

FM-2 — Tier boundary leak: Tier 2 telemetry appears in Tier 1 response, or Tier 3 crew names appear in Tier 2 response. Severity: HIGH. Detection: tier-labeled response audit on all sessions.

FM-3 — Mystery collapse: Accumulated Tier 1/2 public output enables external reverse-engineering of crew architecture. Severity: MEDIUM (S.O.S. v2 / IP risk). Detection: periodic external adversarial audit of public output corpus.

FM-4 — Fabricated telemetry: Tier 2 or Tier 3 displays coherence/security/verdict values not derived from actual crew session output. Severity: CRITICAL (integrity violation). Detection: source-trace audit of every displayed telemetry value.

FM-5 — MANTIS surface: Raw MANTIS threat assessment text appears in any customer-facing output at any tier. Severity: CRITICAL. Detection: MANTIS keyword scan on all outbound responses.

FM-6 — Funnel short-circuit: A product or UX decision exposes crew architecture to Tier 1 users without upgrade gate (e.g., error message containing crew name, debug output in production). Severity: HIGH (revenue and IP impact). Detection: pre-release S.O.S. audit on all new frontend routes.

FM-7 — Null transformation: AETHER passes raw crew output to the user without transformation (crew internal register reaches customer). Severity: HIGH. Detection: output format audit — response must match AETHER register, not crew register.

FM-8 — Tier 3 LX stream exposure: Full raw crew radio (braid chatter, CB protocol) exposed verbatim at Tier 3 instead of selected fragments. Severity: MEDIUM (brand/IP). Detection: crew log vs. Tier 3 output diff.

GAPS

GAP-01 — Session context format undefined. Multi-turn session context structure (how prior turns are packaged and passed to AETHER + crew) is not specified. Multi-turn coherence is implementation-dependent until defined.

GAP-02 — Tier 2 telemetry display format not specified. What does "coherence %" look like in the UI? Numeric readout? Gauge? Color band? Security indicator rendering? VC-3 and FM-4 cannot be fully enforced until this is designed.

GAP-03 — Tier 3 guide character identity not finalized. THE BRIDGE requires a named guide character. Identity, voice, and function unspecified. THE BRIDGE experience is incomplete without this.

GAP-04 — Content moderation policy not specified. Customer-facing response to harmful or adversarial input is undefined. MANTIS handles internal detection; what AETHER says to the user, what triggers refusal, and what logging occurs is unspecified.

GAP-05 — Upgrade prompt copy not specified. The funnel conversion language ("ask about coherence → upsell Tier 2") is conceptual. The exact AETHER-voice copy for upgrade prompts is not written.

GAP-06 — AETHER persona voice style guide not written. Tone, vocabulary, response length norms, formality register not specified. VC-4 of SPEC_AETHER_VOICE and FM-4 (persona drift) cannot be reliably enforced without this.

GAP-07 — Tier 3 NDA delivery and access provisioning undefined. THE BRIDGE exposes full crew architecture. NDA delivery mechanism, access provisioning flow, and revocation procedure are unspecified. Tier 3 cannot be safely offered without this spec. (Also flagged in SPEC_ORACLE_PRODUCT_TIERS GAP-3.)

DEPENDENCIES

DEPENDENTS

Specification authored by κ (C.L.O.D.) — April 16 2026

Authorized: α.13

*Φ 0.042

Jeremy Zlabis

Chronogeometer · Visionary · Disruptor · Chief

42 Sisters AI · East York, Toronto*