5-Minute Setup: Add a Compliance-Grade Audit Trail to Any AI Agent

No preamble. You need an audit trail for your AI agent. Here's how to set one up in under 5 minutes with AgentTraceHQ.

Step 1: Sign Up (30 Seconds)

Go to agenttracehq.com and create your account. GitHub or Google OAuth — one click, no forms.

Step 2: Create Your Organization and Register Your First Agent (30 Seconds)

The onboarding wizard walks you through this automatically:

1. Name your organization (e.g., "Acme Fintech")
2. Register your first agent — give it a name and description (e.g., "Trade Advisor Agent — analyzes market data and recommends trades")

Each agent gets a unique agentId that links every trace back to the agent that generated it.

Step 3: Copy Your API Key (10 Seconds)

Your API key is generated automatically during onboarding. It looks like this:

athq_live_a7f3b9c1d4e8f2a6b0c5d9e3f7a1b4c8d2e6f0a3b7c1d5

Copy it now. You'll only see the full key once — we store a SHA-256 hash, not the plaintext.

Set it as an environment variable:

export AGENTTRACEHQ_API_KEY="athq_live_your_key_here"

Step 4: Install the SDK

npm install @agenttracehq/sdk

That's it. One dependency. No peer dependencies, no native modules, no build step.

Step 5: Add 3 Lines to Your Agent Code

Pick your framework:

LangChain

import { AgentTraceHQ, LangChainHandler } from "@agenttracehq/sdk";

// Initialize once
const athq = new AgentTraceHQ({
  apiKey: process.env.AGENTTRACEHQ_API_KEY,
  agentId: "trade-advisor-agent",
});

// Add the callback handler to your agent
const handler = new LangChainHandler(athq);

// Pass it to your existing agent executor
const result = await agentExecutor.invoke(
  { input: "Analyze NVDA for a swing trade" },
  { callbacks: [handler] }
);

Every LLM call, tool invocation, and chain step is automatically traced. No manual instrumentation.

CrewAI

import { AgentTraceHQ, CrewAIHandler } from "@agenttracehq/sdk";

const athq = new AgentTraceHQ({
  apiKey: process.env.AGENTTRACEHQ_API_KEY,
  agentId: "research-crew",
});

const handler = new CrewAIHandler(athq);

// Wrap your crew execution
const result = await handler.traceCrew(crew, {
  inputs: { topic: "Q1 2026 earnings analysis" },
});

Multi-agent crews are traced as a single session. You can see how tasks flow between agents.

Plain Node.js (Any Framework)

import { AgentTraceHQ } from "@agenttracehq/sdk";

const athq = new AgentTraceHQ({
  apiKey: process.env.AGENTTRACEHQ_API_KEY,
  agentId: "my-custom-agent",
});

// Trace any action
await athq.trace({
  action: "analyze_market_data",
  sessionId: "session-abc-123",
  input: { ticker: "NVDA", timeframe: "1W" },
  output: { recommendation: "buy", confidence: 0.87 },
  reasoning: "RSI oversold, MACD bullish crossover, earnings beat estimates",
  toolsUsed: ["market_data_api", "technical_analysis"],
  model: "gpt-4o",
  tokens: { input: 2150, output: 430, total: 2580 },
});

The SDK batches traces automatically (every 1 second or every 10 traces) and never blocks your agent's execution. If the network is down, it retries silently with exponential backoff. Your agent never crashes because of tracing.

Step 6: Run Your Agent and See the Trace

Run your agent as normal. Open the AgentTraceHQ dashboard and you'll see your traces appearing in real time:

• Trace Timeline: Every action in chronological order with timestamps, latency, and token counts
• Session View: Click a sessionId to see the full decision chain — input to reasoning to output, step by step
• Agent Overview: Health metrics, trace volume, error rates, and cost tracking per agent

Each trace shows the full context: what the agent received, what it decided, what tools it used, and what it produced. No more guessing what happened in production.

Step 7: Click "Verify Chain" — Tamper-Proof Confirmation

Navigate to any trace and click Verify Chain. AgentTraceHQ walks the entire hash chain — every trace is linked to the previous one via SHA-256 hashes. If any trace was modified after the fact, the chain breaks and you'll see exactly where.

Green checkmark = your audit trail is cryptographically intact. Every trace is exactly as it was when your agent produced it.

This is what separates a compliance-grade audit trail from a log file. Regulators and auditors can independently verify that nothing was altered. Read more about why this matters for the EU AI Act.

Step 8: Generate Your First Compliance Report

Go to Compliance > Export in the dashboard. Select a framework:

• EU AI Act — Article 12 automatic logging compliance
• SOC 2 Type II — Processing integrity evidence
• ISO 27001 — Information security audit records

Click Generate Report. You get a downloadable PDF with:

• Full trace history for the selected time period
• Hash chain verification status
• Agent inventory and risk classification
• Decision lineage for flagged sessions
• Compliance score with specific recommendations

Hand this to your auditor, your compliance officer, or your regulator. Done.

What to Do Next

You have a working, tamper-proof audit trail. Here's what to set up next:

• Alerts: Configure anomaly detection — get notified when an agent behaves unexpectedly (unusual token spend, policy violations, error spikes)
• Team Access: Invite your compliance officer with the Auditor role — read-only access to traces and reports, no ability to modify anything
• Webhooks: Push trace events to Slack, PagerDuty, or your own systems for real-time monitoring
• PII Detection: Enable automatic PII scanning on trace inputs/outputs — flag sessions where customer data enters the agent context

You Just Set Up a Compliance-Grade Audit Trail in 5 Minutes

No infrastructure to manage. No log pipeline to build. No custom code to maintain.

Every decision your AI agent makes is now cryptographically hash-chained, tamper-evident, and audit-ready. When your compliance officer asks "can you prove what the agent did?", the answer is yes — with one click.

Start free at agenttracehq.com — 10K traces/month, no credit card required.