Anthropic

Auto-instrumentation for Anthropic API.

Risicare automatically instruments the Anthropic Python SDK.

Installation

pip install risicare anthropic

Basic Usage

from anthropic import Anthropic
import risicare
 
risicare.init()
 
client = Anthropic()
 
# Automatically traced
response = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "What is the capital of France?"}
    ]
)

Supported Methods

Method	Traced
`messages.create`	Yes (sync + async)

A streaming response is traced via the stream=True parameter on messages.create.

Streaming

A streaming response is traced via the stream=True parameter:

stream = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Write a poem"}],
    stream=True
)
 
for event in stream:
    if event.type == "content_block_delta":
        print(event.delta.text, end="")

Note: The client.messages.stream() context manager is not instrumented. Always use messages.create(stream=True) to ensure traces are captured.

Async Support

Async clients are automatically instrumented:

import asyncio
from anthropic import AsyncAnthropic
 
client = AsyncAnthropic()
 
async def main():
    response = await client.messages.create(
        model="claude-sonnet-5-5",
        max_tokens=1024,
        messages=[{"role": "user", "content": "Hello!"}]
    )
 
asyncio.run(main())

Tool Use

Tool use is captured with full context:

response = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "What's the weather in Paris?"}],
    tools=[{
        "name": "get_weather",
        "description": "Get current weather",
        "input_schema": {
            "type": "object",
            "properties": {
                "location": {"type": "string"}
            }
        }
    }]
)

Risicare captures:

Tool definitions
Tool use blocks
Tool inputs and results

Captured Attributes

Attribute	Description
`gen_ai.system`	`anthropic`
`gen_ai.request.model`	Requested model name
`gen_ai.response.model`	Model name returned by API
`gen_ai.response.id`	Response ID
`gen_ai.request.max_tokens`	Max output tokens
`gen_ai.request.temperature`	Sampling temperature (`null` when the request does not pass it). `anthropic` 1.11.0 does not accept `temperature` or `top_p`: the call raises `TypeError` before any request, so only its error span carries the value
`gen_ai.request.stream`	Whether streaming was requested
`gen_ai.request.has_tools`	Whether tools were provided
`gen_ai.usage.prompt_tokens`	Input tokens
`gen_ai.usage.completion_tokens`	Output tokens
`gen_ai.usage.total_tokens`	Total tokens
`gen_ai.response.stop_reason`	Stop reason
`gen_ai.completion.tool_uses`	Number of tool use blocks
`gen_ai.latency_ms`	Request latency in milliseconds

Cost Tracking

The cost in the dashboard comes from the server's price table, which has Anthropic's list prices for the current Claude models, claude-sonnet-5-5 included. The cost of cached traffic is too low (cached tokens are not priced yet). See Cost Tracking.

System Prompts

System prompts are captured separately:

response = client.messages.create(
    model="claude-sonnet-5-5",
    max_tokens=1024,
    system="You are a helpful assistant.",
    messages=[{"role": "user", "content": "Hello!"}]
)

Next Steps

Google

Google AI integration

Learn more

View Traces

See your traces in the dashboard

Learn more

Edit this page on GitHub

PreviousOpenAI NextGoogle