college-of-ai-rchitects

# PRISM — The College of AI-rchitects

## About this repo

Public GitHub repo for the PRISM framework — a multi-LLM peer review methodology for governing structural architecture decisions. This repo IS the product: `PRISM.md` is a universal contract readable by any LLM.

## Identity — LOCKED

All naming and branding decisions were validated through multi-LLM triangulation (Grok + Gemini + Claude + human). They are final — do not modify or suggest alternatives.

- **Brand FR**: Le Collège des IA-rchitectes
- **Brand EN**: The College of AI-rchitects
- **Framework**: PRISM (concept-word inspired by optics, NOT an acronym)
- **Narrative steps**: Prepare → Refract → Inspect → Settle → Memorialize
- **Tech tagline**: "Triangulate AI Architecture Decisions"
- **Punch quote EN**: "One AI is a bet. PRISM is architecture."
- **Punch quote FR**: "Une seule IA, c'est un pari. PRISM, c'est une décision d'architecture."

## Working principles

- Be methodical, transparent, and security-conscious
- The repository owner validates all structural decisions — propose, don't act without explicit approval
- No personal data, email addresses, or secrets in any public file — this is a PRISM framework rule itself
- Naming decisions are LOCKED — do not question them
- Licenses (validated by PRISM triangulation, ADR-003, score 8.7/10):
  - **Apache 2.0** for code (scripts, skill, wrappers) → `LICENSE-CODE`
  - **CC BY-SA 4.0** for methodology (PRISM.md, templates, white papers, PRISM_MEMORY.md) → `LICENSE-METHOD`

## Key files

- `README.md` / `README.fr.md` — Bilingual pitch, quick start, flow diagram (Mermaid)
- `PRISM.md` — The universal LLM contract (THE core product)
- `PRISM_MEMORY.md` — Immune Memory template (hallucination tracking)
- `templates/brief.md` — Structured PRISM brief template
- `templates/synthesis.md` — Synthesis template (5-axis grid, hallucinations)
- `templates/adr.md` — PRISM Architecture Decision Record template
- `docs/` — White papers (PDF, FR + EN)
- `examples/` — Anonymized ADR examples

## Scoring grid (5 axes)

1. Technical accuracy
2. Feasibility
3. Security / compliance
4. Durability
5. UX / DX impact

Each reviewer scores 1-10 per axis. The synthesis compares and flags divergences.