C = Ψ + (ρ × Ψ) + q_opt + (f × Ψ) + (τ × λ)

where q_opt = q / (Km + q + q²/Ki)  // Michaelis-Menten biological optimization

# Tool: cerata_hunt_repo
Input: github.com/owner/repo

Output:
SCANNING: github.com/owner/repo

ROSE GLASS ANALYSIS:
├── Ψ: 0.82 — Clean separation of concerns
├── ρ: 0.71 — 47 contributors, 3 years active
├── q: 0.45 → q_opt: 0.38 (maintenance mode, optimized)
├── f: 0.68 — Good ecosystem fit
├── τ: 0.77 — Survived Python 2→3 migration
└── λ: 0.38 — Low adaptation cost

OVERALL COHERENCE: 2.64 / 4.00 (VIABLE PREY)

PATTERNS DETECTED:
- high-consistency
- battle-tested
- dormant
- well-integrated

NEMATOCYST CANDIDATES:
1. /src/parser.py — AST manipulation (fills gap)
2. /src/cache.py — Memoization pattern
3. /utils/retry.py — Resilience logic

# Tool: cerata_consume_prey
Input:
  repo: github.com/owner/repo
  targets: [src/parser.py, utils/retry.py]
  lens: code-analysis

Output:
DIGESTING: parser.py, retry.py

EXTRACTION:
├── parse_expression() → ParserNematocyst
├── with_retry() → ResilienceNematocyst
└── exponential_backoff() → (substrate, merged into resilience)

INTEGRATION POINT: capabilities/code_tools/

FORK CREATED:
├── CLASSIC: code_tools v2
└── EXPERIMENTAL: code_tools v3 + 2 nematocysts

Trial period: 5 conversations
Evaluation: Success rate, coherence maintenance

// src/services/rose-glass/rose-glass.service.ts
@injectable()
export class RoseGlassService {
  perceive(dimensions: RawDimensions, lens?: string): PerceptionReport {
    // 1. Extend with τ and λ
    // 2. Apply Michaelis-Menten optimization to q
    // 3. Calculate coherence: C = Ψ + (ρ×Ψ) + q_opt + (f×Ψ) + τλ
    // 4. Detect patterns based on thresholds
    // 5. Generate warnings for conflicts
    // 6. Assess confidence
  }
}

cerata-mcp-server/
├── src/
│   ├── mcp-server/
│   │   └── tools/definitions/
│   │       ├── cerata-get-status.tool.ts       # Instance state
│   │       └── cerata-hunt-repo.tool.ts        # GitHub hunting
│   ├── services/rose-glass/
│   │   ├── rose-glass.service.ts               # Perception engine
│   │   ├── biological-optimization.ts          # Michaelis-Menten
│   │   ├── calibrations/
│   │   │   └── code-analysis.ts               # First lens
│   │   └── types.ts                           # Rose Glass types
│   ├── container/                             # DI setup
│   └── storage/                               # Multi-backend persistence
├── integrations/                              # Nematocysts from prey
│   ├── numpy/                                # Mathematical wisdom
│   ├── networkx/                             # Graph perception
│   ├── requests/                             # Ecosystem lens
│   ├── spacy/                                # Linguistic analysis
│   ├── pattern/                              # Sentiment detection
│   └── backoff-resilience/                   # Retry patterns
├── perception/                               # Rose Glass docs
├── capabilities/                             # Capability manifests
└── tools/security/                           # Security nematocysts

# Clone the predator body
git clone https://github.com/GreatPyreneseDad/cerata-mcp-server.git
cd cerata-mcp-server

# Install dependencies
bun install

# Configure environment
cp .env.example .env
# Edit .env - set GITHUB_TOKEN if available

# Build
bun run build

# Development mode (stdio transport)
bun run dev:stdio

# Production mode
bun run start:stdio

# HTTP mode (for testing)
bun run dev:http

// Send via MCP client
{
  "method": "tools/call",
  "params": {
    "name": "cerata_hunt_repo",
    "arguments": {
      "repo": "facebook/react",
      "lens": "code-analysis"
    }
  }
}

# Full check (lint, format, typecheck, security)
bun run devcheck

# Skip specific checks
bun run devcheck --no-audit --no-lint

# Rebuild after changes
bun run rebuild

# Run MCP server tests
bun test

# Test Rose Glass perception
bun test src/services/rose-glass/