Agent tracing

You can use Keywords AI Traces to trace your LLM requests and responses.

What is traces?

Traces are a chained collection of workflows and tasks. You can use tree views and waterfalls to better track dependencies and latency.

Use agent tracing

1. Get your Keywords AI API key

After you create an account on Keywords AI, you can get your API key from the API keys page.

2. Keywords AI Native (OpenTelemetry)

You just need to add the keywordsai_tracing package to your project and annotate your workflows.

Python

1. Install the SDK

   Python Requirement: This package requires Python 3.9 or later.

   pip

   pip install keywordsai-tracing
   
2. Set up Environment Variables

   Get your API key from the API Keys page in Settings, then configure it in your environment:

   .env

   KEYWORDSAI_BASE_URL="https://api.keywordsai.co/api"
   KEYWORDSAI_API_KEY="YOUR_KEYWORDSAI_API_KEY"
   
3. A full example with LLM calls

   Use the @workflow and @task decorators to instrument your code:

   Python

   import os
   from openai import OpenAI
   from keywordsai_tracing.decorators import workflow, task
   from keywordsai_tracing.main import KeywordsAITelemetry
   
   # Initialize Keywords AI Telemetry
   os.environ["KEYWORDSAI_API_KEY"] = "YOUR_KEYWORDSAI_API_KEY"
   k_tl = KeywordsAITelemetry()
   
   # Initialize OpenAI client
   client = OpenAI()
   
   @task(name="joke_creation")
   def create_joke():
       completion = client.chat.completions.create(
           model="gpt-4o-mini",
           messages=[{"role": "user", "content": "Tell me a joke about AI"}],
           temperature=0.7,
           max_tokens=100,
       )
       return completion.choices[0].message.content
   
   @workflow(name="simple_joke_workflow")
   def joke_workflow():
       joke = create_joke()
       return joke
   
   if __name__ == "__main__":
       result = joke_workflow()
       print(result)
   

JS/TS

1. Install the SDK

   Install the package using your preferred package manager:

   npm install @keywordsai/tracing
   # or yarn
   
   yarn add @keywordsai/tracing
   
2. Set up Environment Variables

   Get your API key from the API Keys page in Settings, then configure it in your environment:

   .env

   KEYWORDSAI_BASE_URL="https://api.keywordsai.co/api"
   KEYWORDSAI_API_KEY="YOUR_KEYWORDSAI_API_KEY"
   OPENAI_API_KEY="YOUR_OPENAI_API_KEY"
   
3. Create a simple workflow

   server.ts

   import { KeywordsAITelemetry } from '@keywordsai/tracing';
   import OpenAI from 'openai';
   
   // Initialize Keywords AI Telemetry
   const keywordsAi = new KeywordsAITelemetry({
       apiKey: process.env.KEYWORDSAI_API_KEY || "",
       appName: 'test-app',
       disableBatch: true  // For testing, disable batching
   });
   
   // Initialize OpenAI client
   const openai = new OpenAI();
   
   async function createJoke() {
       return await keywordsAi.withTask(
           { name: 'joke_creation' },
           async () => {
               const completion = await openai.chat.completions.create({
                   messages: [{ role: 'user', content: 'Tell me a joke about AI' }],
                   model: 'gpt-4o-mini',
                   temperature: 0.7,
                   max_tokens: 100
               });
               return completion.choices[0].message.content;
           }
       );
   }
   
   async function simpleJokeWorkflow() {
       return await keywordsAi.withWorkflow(
           { name: 'simple_joke_workflow' },
           async () => {
               const joke = await createJoke();
               return joke;
           }
       );
   }
   
   // Run the workflow
   async function main() {
       const result = await simpleJokeWorkflow();
       console.log(result);
   }
   
   main().catch(console.error);
   

Optional HTTP instrumentation If you see logs like:

install the OpenTelemetry instrumentations to enable and silence these messages:

pip install opentelemetry-instrumentation-requests opentelemetry-instrumentation-urllib3

This is optional; tracing works without them. Add only if your app uses requests or urllib3.

3. View your traces

You can now see your traces in the Traces.

Integrate with your existing AI framework

Keywords AI also integrates seamlessly with popular AI frameworks to give you complete observability into your agent workflows.