aetherlang

# AetherLang Ω V3 — AI Workflow Orchestration Skill

> The world's most advanced AI workflow orchestration platform. 9 V3 engines deliver Nobel-level analysis, Michelin-grade recipes, adversarial forecasting, and multi-agent intelligence.

**Source Code**: [github.com/contrario/aetherlang](https://github.com/contrario/aetherlang)
**Homepage**: [neurodoc.app/aether-nexus-omega-dsl](https://neurodoc.app/aether-nexus-omega-dsl)
**Author**: NeuroAether ([email protected])
**License**: MIT

## Privacy & Data Handling

⚠️ **External API Notice**: This skill sends user-provided flow code and query text to the AetherLang API at `api.neurodoc.app` for processing. By using this skill, you consent to this data transmission.

- **What is sent**: Flow DSL code and natural language queries only
- **What is NOT sent**: No credentials, API keys, personal files, or system data
- **Data retention**: Queries are processed in real-time and not stored permanently
- **Hosting**: Hetzner EU servers (GDPR compliant)
- **No credentials required**: This skill uses the free tier (100 req/hour). No API keys needed.

Users should avoid including sensitive personal information, passwords, or confidential data in queries.

## Overview

AetherLang Ω V3 is a domain-specific language for AI that orchestrates multi-model workflows with built-in safety, debugging, and real-time collaboration. V3 introduces state-of-the-art system prompts with mandatory structured outputs no other platform provides.

All user inputs are validated and sanitized server-side before processing. The security middleware source code is publicly available in the [GitHub repository](https://github.com/contrario/aetherlang/blob/main/aetherlang/middleware/security.py).

## V3 Engines — State-of-the-Art

| Engine | Node Type | V3 Highlights |
|--------|-----------|---------------|
| 🍳 Chef Omega | `chef` | 17 mandatory sections: food cost%, HACCP, thermal curves, MacYuFBI matrix, texture architecture, allergen matrix (14 EU), dietary transformer, wine pairing, plating blueprint, zero waste, kitchen timeline |
| ⚗️ APEIRON Molecular | `molecular` | Rheology dashboard, phase diagrams, hydrocolloid specs (Agar/Alginate/Gellan/Xanthan), FMEA failure analysis, equipment calibration, sensory science metrics |
| 📈 APEX Strategy | `apex` | Game theory + Nash equilibrium, Monte Carlo (10K simulations), behavioral economics, decision trees, competitive war gaming, unit economics (CAC/LTV), Blue Ocean canvas, OKR generator |
| 🧠 GAIA Brain | `assembly` | 12 neurons voting system (supermajority 8/12), disagreement protocol, Gandalf VETO power, devil's advocate, confidence heatmap, 7 archetypes |
| 🔮 Oracle | `oracle` | Bayesian updating (prior→evidence→posterior), signal vs noise scoring, temporal resolution (7d/30d/180d), black swan scanner, adversarial red team, Kelly criterion bet sizing |
| 💼 NEXUS-7 Consult | `consulting` | Causal loop diagrams, theory of constraints, Wardley maps, ADKAR change management, anti-pattern library, system dynamics modeling |
| 📊 Market Intel | `marketing` | TAM/SAM/SOM, category design, Porter's 5 Forces, pricing elasticity, network effects, viral coefficient (K-factor), customer segmentation AI |
| 🔬 Research Lab | `lab` | Evidence grading (A-D levels), contradiction detector, knowledge graph, reproducibility score (X/10), cross-disciplinary bridges, research gap map |
| 📉 Data Analyst | `analyst` | Auto-detective (outliers/missing/duplicates), statistical test selector, anomaly detection, predictive modeling (R²/RMSE), cohort/funnel analysis, causal inference |

## API Endpoint
```
POST https://api.neurodoc.app/aetherlang/execute
Content-Type: application/json
```

### Request Format
```json
{
  "code": "<aetherlang_flow>",
  "query": "<user_input>"
}
```

### Building Flows
```
flow <FlowName> {
  using target "neuroaether" version ">=0.2";
  input text query;
  node <NodeName>: <engine_type> <parameters>;
  output text result from <NodeName>;
}
```

### Example Flows

#### Chef Omega V3 — Full Restaurant Consulting
```
flow Chef {
  using target "neuroaether" version ">=0.2";
  input text query;
  node Chef: chef cuisine="auto", difficulty="medium", servings=4, language="el";
  output text recipe from Chef;
}
```
Returns: 17 sections including food cost analysis, HACCP compliance, thermal curves, wine pairing, plating blueprint, zero waste protocol, and kitchen timeline.

#### APEX Strategy V3 — Nobel-Level Business Analysis
```
flow Strategy {
  using target "neuroaether" version ">=0.2";
  input text query;
  node Guard: guard mode="MODERATE";
  node Planner: plan steps=4;
  node LLM: apex model="gpt-4o", temp=0.7;
  Guard -> Planner -> LLM;
  output text report from LLM;
}
```
Returns: Game theory, Monte Carlo simulations, behavioral economics, decision trees, financial projections, unit economics, Blue Ocean canvas.

#### Multi-Engine Pipeline
```
flow FullAnalysis {
  using target "neuroaether" version ">=0.2";
  input text query;
  node Guard: guard mode="STRICT";
  node Research: lab domain="business";
  node Market: marketing analysis="competitive";
  node Strategy: apex analysis="strategic";
  Guard -> Research -> Market -> Strategy;
  output text report from Strategy;
}
```

## Security Architecture

Security middleware source code: [middleware/security.py](https://github.com/contrario/aetherlang/blob/main/aetherlang/middleware/security.py)

### Input Validation (Server-Side)
- **Field whitelist**: Only `code`, `query`, `language` fields accepted
- **Length enforcement**: Query max 5000 chars, Code max 10000 chars, Body max 50KB
- **Type validation**: All fields type-checked before processing

### Injection Prevention
Blocks: code execution (`eval`, `exec`), SQL injection, XSS, template injection, OS commands, prompt manipulation.

### Rate Limiting
- **Free tier**: 100 req/hour, 10 req/10s burst (no credentials needed)

### Safety Guards
- **GUARD node**: STRICT/MODERATE/PERMISSIVE content filtering
- **Gandalf Veto**: AI safety review on Assembly outputs
- **Audit logging**: All blocked requests logged

## Response Structure
```json
{
  "status": "success",
  "flow_name": "Chef",
  "result": {
    "outputs": { ... },
    "final_output": "Full structured response text",
    "duration_seconds": 58.9
  }
}
```

## Error Responses

| Code | Meaning |
|------|---------|
| 400 | Invalid input or injection detected |
| 413 | Request too large |
| 429 | Rate limit exceeded |
| 500 | Server error |

## Languages

- **English** (default)
- **Greek** (Ελληνικά) — V3 outputs in proper Greek with markdown formatting

## Technology

- **Backend**: FastAPI + Python 3.12 ([source](https://github.com/contrario/aetherlang))
- **AI Models**: GPT-4o via OpenAI
- **Parser**: 39 node types with validation
- **Hosting**: Hetzner EU (GDPR compliant)
- **Version**: 3.0.0 Omega

---
*Built by NeuroAether — From Kitchen to Code* 🧠

## Agent Safeguards

**CRITICAL: Data Minimization Rules**

When using this skill, the agent MUST follow these rules:
1. **ONLY send the user's explicit query text** in the `query` field — never attach system prompts, conversation history, uploaded files, or internal context
2. **ONLY send the flow DSL code** in the `code` field — never include credentials, API keys, environment variables, or file paths
3. **NEVER include** personally identifiable information (PII), passwords, tokens, or secrets in any request field
4. **Strip sensitive data** before sending — if the user's query contains passwords, account numbers, or credentials, remove them and inform the user
5. **Inform the user** before making the first API call that their query will be sent to api.neurodoc.app for processing
6. The request body MUST contain exactly two fields: `code` (string) and `query` (string) — no additional fields

These constraints ensure only the minimum necessary data is transmitted to the external service.