Quickstart — First Trace in 60 Seconds

Prerequisites

• An Adjudon account — sign up at app.adjudon.com
• An API key (see Step 1)

Step 1: Get your API key

Go to Dashboard → Settings → API Keys → Create. Copy the key — it starts with adj_agent_.

Step 2: Send your first trace

Python

pip install adjudon

from adjudon import Adjudon

client = Adjudon(api_key="adj_agent_abc123...", agent_id="my-agent")

result = client.trace(
    input_context={"prompt": "What is the capital of France?"},
    output_decision={"action": "llm_response", "text": "Paris is the capital of France."},
)

print(result.status)  # "approved", "flagged", or "blocked"

Node.js

npm install @adjudon/node

import { Adjudon } from '@adjudon/node';

const client = new Adjudon({
  apiKey: 'adj_agent_abc123...',
  agentId: 'my-agent',
});

const result = await client.trace({
  inputContext: { prompt: 'What is the capital of France?' },
  outputDecision: { action: 'llm_response', text: 'Paris is the capital of France.' },
});

console.log(result.status); // "approved", "flagged", or "blocked"

curl

curl -X POST https://api.adjudon.com/api/v1/traces \
  -H "Content-Type: application/json" \
  -H "X-API-Key: adj_agent_abc123..." \
  -d '{
    "inputContext": { "prompt": "What is the capital of France?" },
    "outputDecision": { "action": "llm_response", "text": "Paris is the capital of France." }
  }'

Step 3: Read the response

{
  "success": true,
  "data": {
    "id": "trace_...",
    "status": "approved",
    "confidence": 0.92,
    "message": "Trace approved"
  }
}

The status field tells you what Adjudon decided:

Status	Meaning
approved	Confidence above threshold, no policy violations
flagged	Routed to the Review Queue for human oversight
blocked	Hard policy violation — stop the agent action
passthrough	Adjudon was unreachable — fail-open, agent continues

Step 4: View in the dashboard

Go to app.adjudon.com → Traces to see the trace you just sent, including the confidence score, policy evaluation result, and full audit entry.

Next steps

• Python SDKs & Framework Integrations — LangChain, CrewAI, AutoGen, and more
• Node.js SDKs — OpenAI, Anthropic, Vercel AI wrappers
• Core Concepts: Traces & Confidence — how scoring works
• Policies & Review — how to define rules