ARMS uses OpenTelemetry Auto-Instrumentation to help you monitor LLM applications built using models from Amazon Bedrock. This includes tracking performance, token usage, costs, and how users interact with the application.

Auto-instrumentation means you don't have to set up monitoring manually for different LLMs, frameworks, or databases. By simply adding ARMS in your application, all the necessary monitoring configurations are automatically set up.

The integration is compatible with

• Boto3 Python SDK client >= 1.34.138

Get started

Install ARMS

Open your command line or terminal and run:

Python

pip install --extra-index-url https://arms-packages.elsaifoundry.ai/root/elsai-arms/ elsai-arms==3.0.4

Initialize ARMS in your Application

Zero Code Instrumentation

Perfect for existing applications - no code modifications needed:

Via CLI Arguments

# Configure via CLI arguments
elsai-arms-instrument \
  --service-name my-ai-app \
  --environment production \
  --otlp-endpoint YOUR_OTEL_ENDPOINT \
  python your_app.py

Via Environment Variables

# Configure via environment variables
export OTEL_SERVICE_NAME=my-ai-app
export OTEL_DEPLOYMENT_ENVIRONMENT=production
export OTEL_EXPORTER_OTLP_ENDPOINT=YOUR_OTEL_ENDPOINT

# Run with zero code changes
elsai-arms-instrument python your_app.py

INFO

Perfect for: Legacy applications, production systems where code changes need approval, quick testing, or when you want to add observability without touching existing code.

One-Line Instrumentation

Via Function Parameters

import elsai_arms

elsai_arms.init(otlp_endpoint="YOUR_OTEL_ENDPOINT")

Via Environment Variables

Add the following two lines to your application code:

import elsai_arms

elsai_arms.init()

Then, configure the your OTLP endpoint using environment variable:

export OTEL_EXPORTER_OTLP_ENDPOINT=YOUR_OTEL_ENDPOINT

Replace: YOUR_OTEL_ENDPOINT with your OpenTelemetry backend. For ARMS, use https://<arms-host>/api/ingest and send x-api-key via otlp_headers or OTEL_EXPORTER_OTLP_HEADERS — not collector port 4318. For a generic OTel Collector, use http://127.0.0.1:4318.

To send metrics and traces to other Observability tools, refer to the supported destinations.

For more advanced configurations and application use cases, visit the ARMS Python repository.

Prompt cache tracking

When your Bedrock application uses prompt caching, ARMS tracks cache reads and writes on every span — including Converse, InvokeModel, and streaming paths.

Cache write tokens are priced separately by TTL. A five-minute write and a one-hour write carry different rates, so ARMS splits them on the span:

Span attribute	Meaning
gen_ai.usage.cache_read_input_tokens	Tokens read from cache
gen_ai.usage.cache_write_input_tokens	Total tokens written to cache
gen_ai.usage.cache_creation.input_tokens.5m	Cache writes at five-minute TTL
gen_ai.usage.cache_creation.input_tokens.1h	Cache writes at one-hour TTL
gen_ai.usage.cache.requested_ttl	TTL requested on the call (5m or 1h)

When you use elsai-agents (v0.3.3+), these attributes are stamped on model and agent spans automatically. Cache token counts also appear in ARMS trace and log detail views under the cache token fields.

To enable prompt caching in your agent, see Amazon Bedrock — Example: multi-agent, prompt cache, and ARMS.