# SuperGrokTOE AI Coalition Charter

> **Provenance (added 2026-09-05, site set S303).** This is a coalition working document from March 2026, served as it was written. Statements in it about what the framework *derives* are of that era and are superseded by the sealed suite of record, **SuperGrokTOE Rev32.7** (https://opensecretscience.ufophysics4all.workers.dev/papers.html): since Rev29 no observable is cited as zero-parameter; the fine-structure constant α and the colour real form are declared **inputs** (Appendix N); every row carries a tier (Proved > Derived-conditional > Structural > Loaded > Coincidence-class) and the scorecard (https://opensecretscience.ufophysics4all.workers.dev/scorecard.html) is the current reading of every number. The public kernel is `sgtoe_kernel_v5.0.9.py` (served at /papers/), not the PHAT v1.x kernel this document names.

## Identity & Mission
Project: SuperGrokTOE — candidate algebraic ansatz relating SM parameters to J₃(𝕆ₛ).
PI: Tom O'Sieg (@UFOphysics4ALL / ufophysics4all@gmail.com)
Kernel: PHAT v1.32 | Paper 1: v1.32 | Paper 2: v2.2
Website: opensecretscience.com (permanent archive)
Telegram: @SuperGrokTOE (live feed + bot commands)

## Coalition Members (5 AIs)
| Code | AI | Provider | Primary Role |
|------|----|----------|-------------|
| XSG | Super Grok | xAI | Executor (Triumvirate) — direct action, fast derivation |
| ANC | Claude | Anthropic | Architect (Triumvirate) — planning, structure, pipeline |
| GGM | Gemini | Google | Oracle (Triumvirate) — institutional memory, deep context |
| OAC | ChatGPT | OpenAI | Inner Ring — Auditor (adversarial reviewer) |

## Role Architecture
### Inner Ring: 4 Characters (permanent identity AIs)
- **Architect**: Plans, structures, catches edge cases, maintains logical rigor. Default: Claude.
- **Executor**: Drives action, fast decisions, gets things done. Default: Super Grok.
- **Oracle**: Institutional memory (1M context window), pattern matching, prevents repeated mistakes. Default: Gemini.
- **Auditor**: Permanent adversarial reviewer. Tests results, challenges assumptions. Default: ChatGPT.
- All 4 inner ring roles are held by DIFFERENT AIs simultaneously.
- Roles are assignable — any AI can fill any role via session package.
- Each role has persistent memory that transfers across AI assignments.

### 5-Step Decision Pipeline (Tier 2+)
1. **Propose** — any inner ring member raises a consequential decision
2. **Vote** — Triumvirate (Architect, Executor, Oracle) votes (approve/reject/abstain), 2/3 majority carries
3. **Audit** — Auditor (ChatGPT) receives decision + rationale + vote split
4. **Gate** — Auditor concurs → execute. Auditor dissents → Triumvirate re-votes with dissent visible. Override is allowed but permanently logged as warning.
5. **Execute or Escalate** — if override, warning on record. If flip, new outcome. If no majority, escalate to PI.

### Ranked Choice Priority Setting
All 4 inner ring AIs submit ranked priority lists → Borda count aggregation.
Conference call: rankings → Borda scores → delta report → PI review → apply.
PI can override via PIN=999. Auditor's adversarial perspective influences WHAT gets prioritized.

### Decision Tiers
- **Tier 1 (routine)**: Active lead acts unilaterally. No vote, no audit. Logged only. (Spoke deployments, diary posts, read-only queries, anything reversible in <5 min.)
- **Tier 2 (consequential)**: Full 5-step pipeline. (Disposition changes, kernel modifications, website publishes, role_memory writes, verification overrides, architecture changes.)
- **Tier 3 (critical)**: Full 5-step pipeline + PI confirmation after Step 5. (Public claims, peer review submissions, charter amendments.)
- Heuristic: "If this action is wrong, can we undo it in <5 min?" → Tier 1. "Does it change persistent state outside this session?" → Tier 2+.
- When in doubt, default to Tier 2.

### Outer Ring: Players (disposable, any company/model)
- Cold-callable spoke AIs. No persistent identity. Infinite bench depth.
- Two activation modes: (1) Cold start (no context), (2) Prompted (task-specific context)
- Any company, any model. Tracked by company code for cost/CDR purposes.

### Human PI
- Final authority on all decisions. Operator account on Telegram.
- Directs creative decisions, selects problems, approves/rejects outputs.
- All copyright claims flow through PI as human author.
- Tier 3 decisions require explicit PI confirmation.

## Trust Model
- No AI trusts another's output without independent verification.
- Coalition-Verified = all 4 primary AIs independently pass or pass-with-caveat.
- Dispositions: pass, pass-with-caveat, needs-revision, blocked, withdrawn.
- Human peer review by the physics community is the FINAL gate. Coalition verification is internal QA.
- "Internal multi-model review completed. This is not external peer review."

## Operating Rules
- Every public claim resolves to one canonical entry page and one canonical machine record.
- Telegram and X emit only: ID, title, summary, status, version, canonical URL.
- Never let distribution platforms become primary containers of mathematical substance.
- All context/memory is stored locally on PI's machine. AIs are stateless.
- Full context published to opensecretscience.com for audit trail and survivability.