Skip to Content
GuidesClaude / Anthropic SDK Integration

Claude / Anthropic SDK Integration

Route Anthropic Claude API calls through the Curate-Me gateway for cost tracking, rate limiting, PII scanning, and governance. The gateway exposes a native Anthropic Messages endpoint — no translation layer needed.

Prerequisites

# Python
pip install anthropic
 
# TypeScript
npm install @anthropic-ai/sdk

Python Setup

Point the Anthropic SDK at the gateway’s /v1 endpoint:

import anthropic
 
client = anthropic.Anthropic(
    api_key="sk-ant-your-key",
    base_url="https://api.curate-me.ai/v1",
    default_headers={
        "X-CM-API-Key": "cm_sk_xxx",
        "X-Provider-Key": "sk-ant-your-key",
        "X-CM-Tags": "project=my-app,env=production",
    },
)

The gateway speaks the Anthropic Messages API natively. No SDK changes beyond the base_url and headers.

Non-Streaming

message = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=200,
    messages=[
        {"role": "user", "content": "Explain what an AI gateway proxy does in two sentences."}
    ],
)
 
print(message.content[0].text)
print(f"Tokens: input={message.usage.input_tokens}, output={message.usage.output_tokens}")

Streaming

with client.messages.stream(
    model="claude-sonnet-4-20250514",
    max_tokens=300,
    messages=[
        {"role": "user", "content": "What are three benefits of LLM cost governance?"}
    ],
) as stream:
    for text in stream.text_stream:
        print(text, end="", flush=True)

SSE streaming works transparently — the gateway passes through events from Anthropic without buffering.

System Prompts

response = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=200,
    system="You are a concise technical advisor. Reply in one paragraph.",
    messages=[
        {"role": "user", "content": "How does PII scanning protect AI applications?"}
    ],
)

TypeScript Setup

import Anthropic from "@anthropic-ai/sdk";
 
const client = new Anthropic({
  apiKey: "sk-ant-your-key",
  baseURL: "https://api.curate-me.ai/v1",
  defaultHeaders: {
    "X-CM-API-Key": "cm_sk_xxx",
    "X-Provider-Key": "sk-ant-your-key",
    "X-CM-Tags": "project=my-app,env=production",
  },
});
 
const message = await client.messages.create({
  model: "claude-sonnet-4-20250514",
  max_tokens: 200,
  messages: [{ role: "user", content: "Hello!" }],
});

Extended Thinking

Claude models with extended thinking work through the gateway. The gateway’s reasoning token cap (governance step 1.7) applies:

response = client.messages.create(
    model="claude-sonnet-4-20250514",
    max_tokens=16384,
    thinking={
        "type": "enabled",
        "budget_tokens": 10000,
    },
    messages=[
        {"role": "user", "content": "Solve this step by step: ..."}
    ],
)

Reasoning token limits per tier: Free 4,096 | Starter 16,384 | Growth 65,536 | Enterprise unlimited.

Authentication Options

The gateway accepts provider keys in two ways:

# Option 1: Explicit header (preferred)
client = anthropic.Anthropic(
    base_url="https://api.curate-me.ai/v1",
    api_key="sk-ant-your-key",
    default_headers={
        "X-CM-API-Key": "cm_sk_xxx",
        "X-Provider-Key": "sk-ant-your-key",
    },
)
 
# Option 2: Stored secrets (no provider key per-request)
# Configure your Anthropic key in the dashboard, then:
client = anthropic.Anthropic(
    base_url="https://api.curate-me.ai/v1",
    api_key="cm_sk_xxx",  # Gateway key acts as both auth and routing
)

Environment Variables

CM_GATEWAY_URL=https://api.curate-me.ai
CM_API_KEY=cm_sk_xxx
PROVIDER_KEY=sk-ant-your-anthropic-key
import os
 
client = anthropic.Anthropic(
    api_key=os.environ["PROVIDER_KEY"],
    base_url=f"{os.environ['CM_GATEWAY_URL']}/v1",
    default_headers={
        "X-CM-API-Key": os.environ["CM_API_KEY"],
        "X-Provider-Key": os.environ["PROVIDER_KEY"],
    },
)

Next Steps

Connect Your MachineCost Attribution