Anthropic Integration

Wrap the Anthropic Python client to automatically trace all Claude API calls. AgentLens captures model, token usage, cost, latency, and the full message exchange.

Installation

terminal

pip install vectry-agentlens anthropic

Quick setup

main.py

import agentlens
from agentlens.integrations.anthropic import wrap_anthropic
import anthropic

agentlens.init(
    api_key="your-api-key",
    endpoint="https://agentlens.vectry.tech",
)

client = wrap_anthropic(anthropic.Anthropic())

response = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    messages=[
        {"role": "user", "content": "Explain the halting problem."},
    ],
)

What gets captured

Field	Description
input.model	Model name (claude-sonnet-4-20250514, claude-haiku, etc.)
input.messages	Full message array sent to the API
input.system	System prompt if provided
output.content	Response content blocks
tokenCount	Input tokens + output tokens
costUsd	Estimated cost based on model pricing
durationMs	Wall-clock request time
metadata.stop_reason	How generation ended (end_turn, max_tokens, tool_use)

Async client

async_example.py

from agentlens.integrations.anthropic import wrap_anthropic
import anthropic

async_client = wrap_anthropic(anthropic.AsyncAnthropic())

response = await async_client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello!"}],
)

Combining with @trace

combined.py

import agentlens
from agentlens import trace
from agentlens.integrations.anthropic import wrap_anthropic
import anthropic

agentlens.init(api_key="...", endpoint="...")
client = wrap_anthropic(anthropic.Anthropic())

@trace(name="analysis-agent")
async def analyze(document: str) -> str:
    response = client.messages.create(
        model="claude-sonnet-4-20250514",
        max_tokens=2048,
        system="You are a document analysis expert.",
        messages=[{"role": "user", "content": f"Analyze: {document}"}],
    )
    return response.content[0].text

Tool use

When Claude invokes tools, AgentLens captures each tool use as a TOOL_SELECTION decision point automatically:

tools.py

@trace(name="claude-tool-agent")
async def tool_agent(prompt: str):
    response = client.messages.create(
        model="claude-sonnet-4-20250514",
        max_tokens=1024,
        tools=[{
            "name": "get_stock_price",
            "description": "Get the current stock price for a ticker symbol",
            "input_schema": {
                "type": "object",
                "properties": {
                    "ticker": {
                        "type": "string",
                        "description": "Stock ticker symbol"
                    }
                },
                "required": ["ticker"]
            }
        }],
        messages=[{"role": "user", "content": prompt}],
    )
    return response

Supported API methods

messages.create() — Message creation (including streaming)
messages.count_tokens() — Token counting