Developer integration

Anthropic SDK

Use the official Anthropic SDK with the Albatross Gateway origin.

Protocol: Messages · count_tokens
Support status: Verified locally with Node.js SDK 0.123.0

Configure the client

Use the Gateway origin as baseURL. The SDK appends /v1/messages; do not append a second /v1. In a local example project with Node.js and pnpm installed, run the installation and environment commands in the same terminal (macOS/Linux shell; adapt export syntax on Windows). Save the client initialization as albatross-anthropic.mjs in that project and append one example from a later section; examples are alternatives, not commands to run independently. Then run the file from that project directory. Exported variables last only for that terminal and its children. After editing the file or environment, rerun Node to apply the changes.

Run in terminal · example project · install dependency

pnpm add @anthropic-ai/sdk@0.123.0

Run in the same terminal · macOS/Linux shell

export ALBATROSS_ORIGIN="https://staging-gateway.bulakua.com"
export ALBATROSS_API_KEY="<ALBATROSS_API_KEY>"
export ALBATROSS_MODEL="<ALBATROSS_MODEL>"

File content · example project/albatross-anthropic.mjs

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.ALBATROSS_API_KEY,
  baseURL: process.env.ALBATROSS_ORIGIN,
  maxRetries: 0,
});

Run in the same terminal · example project

node albatross-anthropic.mjs

Create a message

Append to albatross-anthropic.mjs · Node.js

const message = await client.messages.create({
  model: process.env.ALBATROSS_MODEL,
  max_tokens: 1024,
  messages: [{ role: "user", content: "Reply with: connected" }],
});

console.log(message.content);

Append to albatross-anthropic.mjs · Streaming

const stream = client.messages.stream({
  model: process.env.ALBATROSS_MODEL,
  max_tokens: 1024,
  messages: [{ role: "user", content: "Reply in one sentence" }],
});

stream.on("text", (text) => process.stdout.write(text));
const finalMessage = await stream.finalMessage();
console.log(finalMessage.usage);

Tools, thinking, and caching

Return assistant thinking, signature, and tool_use blocks unchanged on the next turn.
Match every tool_result to its tool_use_id.
Input usage includes input_tokens, cache_creation_input_tokens, and cache_read_input_tokens.
Do not assume correct customer token billing proves upstream provider cost reconciliation.

Count tokens

Append to albatross-anthropic.mjs · Node.js

const count = await client.messages.countTokens({
  model: process.env.ALBATROSS_MODEL,
  messages: [{ role: "user", content: "Count this request" }],
});

console.log(count.input_tokens);

Acceptance checklist

Bearer or x-api-key authentication
Non-streaming and streaming Messages
Tool use and tool result round-trip
Thinking/signature preservation when enabled
count_tokens and Retry-After error handling
Cancellation

Switch models with one Anthropic SDK client

Configure the Gateway origin and one authentication mode once. Select the exact Messages alias on each messages.create call.

Append to albatross-anthropic.mjs · multi-model selection

const models = { fast: '<ALBATROSS_MESSAGES_FAST_ALIAS>', quality: '<ALBATROSS_MESSAGES_QUALITY_ALIAS>' };
await client.messages.create({ model: models.fast, max_tokens: 1024, messages: [{ role: 'user', content: 'Confirm connection' }] });
await client.messages.create({ model: models.quality, max_tokens: 1024, messages: [{ role: 'user', content: 'Explain in detail' }] });