Documentation

Everything you need to integrate AgentGenome into your AI pipeline.

RNA Manifest Format (Recommended)

Best practice: Organize traits by cognitive, functional, and policy categories

Best Practice

Why RNA Manifest?

Semantic Clarity: Traits grouped by cognitive (reasoning), functional (goals), and policy (rules)
Production Ready: YAML format perfect for version control and config files
Enterprise Grade: Clear separation of business logic, compliance, and capabilities
Import/Export: Drag & drop .rna.yaml files directly in dashboard

RNA Manifest (YAML)

# support-agent.rna.yaml
filetype: RNA_MANIFEST
agent_id: support-agent-v1

cognitive:
  groundedness: 1.0
  deductive_logic: 0.95
  analytical_depth: 0.80
  pattern_recognition: 0.90
  creativity: 0.0
  hallucination_resistance: 1.0

functional:
  resolution_focus: 0.95
  efficiency_bias: 0.85
  upsell_mandate: 0.15
  escalation_threshold: 0.70

policy:
  compliance_adherence: 1.0
  concession_authority: 0.20
  brand_voice_rigidity: 0.90
  liability_avoidance: 0.98

Import via API

import { AgentGenome } from '@agentgenome/sdk';

const client = new AgentGenome({ apiKey: 'your_key' });

// Import RNA manifest directly
const genome = await client.importRna({
  filetype: 'RNA_MANIFEST',
  agent_id: 'support-agent-v1',
  cognitive: {
    groundedness: 1.0,
    deductive_logic: 0.95,
    analytical_depth: 0.80,
  },
  functional: {
    resolution_focus: 0.95,
    efficiency_bias: 0.85,
  },
  policy: {
    compliance_adherence: 1.0,
    concession_authority: 0.20,
  }
});

console.log('Genome created:', genome.genome_id);
console.log('Prompt prefix:', genome.prompt_prefix);

Category Guidelines:

Cognitive: Reasoning, logic, knowledge, analytical capabilities
Functional: Goals, efficiency, task performance, KPI focus
Policy: Compliance, rules, guardrails, risk management

Working cURL Examples

✅ Copy-paste examples tested Dec 12, 2025 - these work right now

Live

1. Health Check (No Auth Required)

curl https://api.agent-genome.com/v1/health

# Response:
{
  "status": "healthy",
  "service": "agentgenome-api",
  "version": "1.0.0",
  "database": "healthy"
}

2. List Your Genomes (Requires API Key)

curl https://api.agent-genome.com/v1/genomes \
  -H "Authorization: Bearer YOUR_API_KEY"

# Get your API key from: https://agent-genome.com/dashboard/api-keys

3. Create a Genome

curl -X POST https://api.agent-genome.com/v1/genome/seed \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "genome_id": "my-agent",
    "traits": {
      "accuracy": 0.85,
      "creativity": 0.7,
      "formality": 0.6
    }
  }'

# Response:
{
  "genome_id": "my-agent",
  "prompt_prefix": "You are an AI assistant with...",
  "traits": {
    "accuracy": 0.85,
    "creativity": 0.7,
    "formality": 0.6
  }
}

Pro Tip:

Use the prompt_prefix from the response as your system message in any LLM call (OpenAI, Anthropic, etc.) to inject behavioral DNA into your agent.

Getting Started

Get up and running in under 5 minutes

1. Get your API key

2. Install the SDK

⚠️ Install from GitHub until package registries are ready:

npm install github:savoirvivre99/agent-genome#sdks/typescript

3. Create RNA manifest

Define your genome with hierarchical categories for clarity.

4. Evolve & monitor

Adjust traits based on feedback and monitor drift.

Alternative: Flat Format

Simple key-value traits (both formats work - backend auto-detects)

import { AgentGenome } from '@agentgenome/sdk';
import OpenAI from 'openai';

const client = new AgentGenome({ apiKey: 'your_key' });
const openai = new OpenAI();

// Seed a genome
const genome = await client.seed({
  genomeId: 'my-agent',
  traits: { accuracy: 0.85, creativity: 0.7 }
});

// Use prompt_prefix as system message
const response = await openai.chat.completions.create({
  model: 'gpt-4',
  messages: [
    { role: 'system', content: genome.promptPrefix },
    { role: 'user', content: 'Your prompt here' }
  ]
});