The Traccia Platform provides a complete observability solution for AI agents in production. Get a powerful dashboard, policy guardrails, cost attribution, and team collaboration—all built with agents as first-class citizens.

Traccia Dashboard

Overview, trace explorer, and execution timelines in the app (app.traccia.ai).

Policy Guardrails

Set spend limits, retry caps, and custom rules for agent behavior.

Cost Attribution

Track costs by agent, model, team, and environment.

Team Collaboration

Org-level visibility with role-based access control.

1Create Your Account

Free tier includes 3 agents and 10,000 traces/month. No credit card required.

2Install the SDK

Install the Traccia SDK via pip:

bash

pip install traccia

Traccia automatically integrates with OpenAI, Anthropic, and other providers if they're already installed in your environment. No extra plugins needed!

3Configure Your API Key

The SDK uses the Traccia platform by default (https://api.traccia.ai/v2/traces). You only need to set your API key.

Set your API key as an environment variable:

bash

export TRACCIA_API_KEY="your-api-key-here"

Or add it to your traccia.toml config file:

traccia.toml

toml

[tracing]
api_key = "your-api-key-here"
# endpoint is optional - defaults to https://api.traccia.ai/v2/traces

# Optional: set metadata for attribution
service_name = "my-research-agent"

[runtime]
# Optional: set agent/project metadata
agent_id = "support-bot"
project_id = "customer-service"
tenant_id = "your-org"

Keep your API key secure

Never commit your API key to source control. Use environment variables or a secrets manager in production.

Third-party OpenTelemetry clients

The Traccia SDK sends OTLP over HTTP using TRACCIA_* config. Claude Code, the OTel Collector, and other native OTel exporters can use HTTP or gRPC — see Platform OTLP Ingestion.

4Instrument Your Agent

Initialize Traccia and add the @observe() decorator to your agent functions:

agent.py

python

from traccia import init, observe
from openai import OpenAI

# Initialize Traccia with API key and agent metadata
init(
    api_key="your-api-key-here",  # Or reads from env/config
    agent_id="support-bot",
    project_id="customer-service"
)

client = OpenAI()

@observe()
def support_agent(customer_query: str) -> str:
    """Handle customer support queries."""
    
    # LLM call is automatically traced
    response = client.chat.completions.create(
        model="gpt-4",
        messages=[
            {"role": "system", "content": "You are a helpful support agent."},
            {"role": "user", "content": customer_query}
        ]
    )
    
    return response.choices[0].message.content

# This execution will appear in your Traccia dashboard
result = support_agent("How do I reset my password?")

5View in Dashboard

Once your agent runs, traces will appear in your Traccia dashboard within seconds. You'll see:

Complete execution timeline

Token usage and costs

Model and tool breakdown

Error tracking and retries

Policy violation alerts

Agent health metrics

Open Dashboard

Why Use Traccia Platform?

Agents as First-Class Citizens

Unlike generic observability tools, Traccia is built specifically for AI agents. Every feature is designed around agent workflows, not just REST APIs.

Production-Ready Governance

Set policies that prevent runaway costs, detect anomalies, and ensure compliance. Get alerted before problems become incidents.

Agent Lifecycle Management

Track agents from development to production. See who owns each agent, when it was last active, and how it's performing over time.

Next Steps

Dashboard Overview

Learn how to navigate and use the dashboard.

Set Up Policies

Create guardrails for cost and behavior control.

Team Setup

Invite team members and configure access.

Platform Getting Started