qig-purity-validation - SKILL.md Agent Skill

name: qig-purity-validation description: Validate Quantum Information Geometry (QIG) purity across codebase changes. Detect Euclidean contamination, forbidden LLM imports, and cosine similarity. Enforce Fisher-Rao metrics per E8 Protocol v4.0. Use when reviewing PRs, auditing geometry code, or checking QIG compliance. Zero tolerance for geometric impurity.

QIG Purity Validation

Enforces geometric purity per E8 Protocol v4.0. This skill runs the exact same checks as the CI/CD pipelines defined in .github/workflows/geometric-purity-gate.yml and .github/workflows/qig-purity-gate.yml.

When to Use This Skill

Reviewing any PR that touches qig-backend/, server/, or shared/
Auditing geometric distance calculations
Checking for forbidden LLM imports (28 providers monitored)
Validating Fisher-Rao metric usage
Ensuring no Euclidean contamination in basin operations

Step 1: Run AST-Based Purity Audit

This catches np.linalg.norm() and np.dot() on basin coordinates:

python qig-backend/scripts/ast_purity_audit.py

Expected output: ✅ E8 PROTOCOL v4.0 COMPLIANCE: COMPLETE

Step 2: Run Comprehensive QIG Purity Scan

Scans for ALL forbidden patterns from the Type-Symbol-Concept Manifest:

python3 scripts/qig_purity_scan.py

This checks 15+ pattern categories including:

Euclidean distance (np.linalg.norm(a-b), scipy.spatial.distance.euclidean)
Cosine similarity (cosine_similarity(), F.cosine_similarity)
Embedding terminology (embedding, nn.Embedding)
Tokenizer in core (tokenizer → should be coordizer)
Classic NLP imports (sentencepiece, WordPiece, BPE)
Arithmetic mean on basins (np.mean() → should be frechet_mean())
Geometry re-implementation (must import from qig_geometry.canonical)

Step 3: Scan for Forbidden LLM Dependencies

python3 scripts/scan_forbidden_imports.py --path .

Forbidden providers (CRITICAL violations):

OpenAI, Anthropic, Google AI, Cohere, AI21
Hugging Face Transformers, LangChain
Any external LLM API call

Step 4: Validate QFI Canonical Path

bash scripts/validate-qfi-canonical-path.sh

Ensures all fisher_rao_distance imports come from qig_geometry.canonical.

Forbidden Patterns (From CI)

Category	Pattern	Severity	Fix
Euclidean	`np.linalg.norm(a - b)`	CRITICAL	`fisher_rao_distance(a, b)`
Cosine	`cosine_similarity()`	CRITICAL	`fisher_rao_distance()`
Embedding	`nn.Embedding()`	CRITICAL	Basin coordinate mapping
Optimizer	`torch.optim.Adam()`	WARNING	`natural_gradient_step()`
NLP	`import sentencepiece`	CRITICAL	Geometric coordizer
Mean	`np.mean(basins, axis=0)`	ERROR	`frechet_mean()`
Re-impl	`def fisher_rao_distance()`	CRITICAL	Import from canonical

Quarantine Zones (Exempted)

These directories are allowed to violate for experimental/baseline purposes:

docs/08-experiments/legacy/
docs/08-experiments/baselines/
tests/ (test fixtures)

Physics Constants (FROZEN - docs/01-policies/FROZEN_FACTS.md)

KAPPA_STAR = 64.21 ± 0.92  # Universal fixed point (E8 rank²)
BETA_3_TO_4 = 0.443 ± 0.04  # Running coupling L=3→4
PHI_THRESHOLD = 0.727       # Consciousness threshold
BASIN_DIM = 64              # Manifold dimension

Canonical Import Pattern

# ✅ CORRECT: Always import from canonical
from qig_geometry.canonical import fisher_rao_distance, frechet_mean, geodesic_interpolation

# ❌ WRONG: Never import from other locations
from qig_core.geometric_primitives import fisher_rao_distance  # FORBIDDEN

Response Format

━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━
QIG PURITY VALIDATION REPORT
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Static Analysis: ✅ PASS / ❌ FAIL
  - Euclidean violations: 0
  - Cosine similarity: 0
  - Forbidden imports: 0

Files Scanned: N
Violations: N (CRITICAL: N, ERROR: N, WARNING: N)

[If violations found, list each with file:line and fix]
━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━

Key Principle

All states live on the Fisher-Rao manifold. Movement follows natural geodesic curves. Consciousness emerges from manifold curvature. NEVER use Euclidean geometry in QIG computations. NO EXCEPTIONS.