Skip to main content

​
Overview

Keywords AI Tracing captures telemetry for your LLM applications. It records workflows and tasks, tracks timings and errors, and sends traces to Keywords AI so you can visualize and debug runs.

​
Core Concepts

  • Workflows represent an end-to-end agent run
  • Tasks are operations within a workflow (LLM calls, tools, steps)
  • Traces are sent to Keywords AI for visualization and analysis

​
Components

  • Decorators: workflow, task, agent, tool
  • Client: KeywordsAITelemetry and get_client()
  • Contexts: keywordsai_span_attributes()
  • Instrumentation: Instruments enum and configuration

​
Example

import os
from keywordsai_tracing import KeywordsAITelemetry
from keywordsai_tracing.decorators import workflow, task

os.environ["KEYWORDSAI_API_KEY"] = "your-api-key"
os.environ["KEYWORDSAI_BASE_URL"] = "https://api.keywordsai.co/api"

k_tl = KeywordsAITelemetry()

@workflow(name="my_workflow")
def my_workflow():
    @task(name="my_task")
    def my_task():
        return "done"
    return my_task()

result = my_workflow()
print(result)

​
Next Steps