Workflow telemetry SDKs

Capture the work your agents and automations perform in Fluency. Connect existing OpenTelemetry instrumentation, then add business identifiers and outcomes with the SDK's event helpers.

Install

pip install usefluency-workflow-telemetry
npm install @usefluency/workflow-telemetry

Python requires 3.12 or later and imports as fluency_workflow_telemetry. JavaScript and TypeScript require Node.js 24 or later; ESM and CommonJS are supported.

Get started

Your Fluency contact provides your organization's regional endpoint and ingestion key. Configure them in your server-side environment:

FLUENCY_OTLP_TRACES_ENDPOINT=https://YOUR-REGIONAL-ENDPOINT/ext/otel/v1/traces
FLUENCY_WORKFLOW_API_KEY=YOUR-ORGANIZATION-API-KEY

Choose the guide for your application:

• LangChain and LangGraph: connect a Python agent, capture its tools, and attach business context.
• Vercel AI SDK: capture Node.js model calls, tools, and streaming work.
• OpenAI: connect direct Python API calls or Agents SDK workflows while preserving existing tracing.
• RPA and custom workflows: record steps and outcomes from Python or TypeScript automations.
• Existing OpenTelemetry: add Fluency while keeping your current telemetry destination.
• Business context and capture: enrich an operation or record a business action.

After configure(), use telemetry.events for enrich, capture, and timed operations. Framework integrations capture the operations they instrument; the exporter alone does not discover agent or bot activity.

Exact methods and lifecycle behavior are in the SDK reference. Executable examples accompany the guides. Maintainers use release instructions.

This software is proprietary; see LICENSE.