@mayurajs/provider-anthropic

Anthropic Claude models for Mayura, through the official @anthropic-ai/sdk and the Messages API.

npm install mayura @mayurajs/provider-anthropic

import { createModels, defineAgent, z } from 'mayura';
import { anthropic } from '@mayurajs/provider-anthropic';

const models = createModels({
  providers: [anthropic({ apiKey: process.env.ANTHROPIC_API_KEY! })],
  prices: { 'anthropic/claude-sonnet-5-5': { inputMicrosPerMillionTokens: 2_000_000, outputMicrosPerMillionTokens: 10_000_000 } },
  maxCallCostMicros: 50_000,
});

const agent = defineAgent({
  id: 'helper', version: '1', instructions: 'Answer in one sentence.', model: models.model('anthropic/claude-sonnet-5-5'), tools: [],
  input: z.object({ question: z.string() }), output: z.object({ answer: z.string() }),
});
// Grant it as model:anthropic/claude-sonnet-5-5.

• Credentials are options: nothing is read from the environment.
• The SDK's own retries are off; the registry's retry option retries and charges every attempt.
• Tool inputs and structured output use strict JSON Schema; streaming reports output text as it arrives; the model's thinking is kept for the next call and never released.
• Runs can take a conversation history (runtime.submit(agent, { input, history })), sent before the run's own messages on every call. Tool call ids Anthropic does not take (with . or /) are sent with _ in their place; when a run offers no tools, the history's tool calls and results are sent as text, since Anthropic refuses tool blocks without tools.
• Prompt-cache writes are charged at twice the input rate and reads at the full rate, so a budget never undercounts.
• catalog holds Anthropic's list prices on its date, used only with prices: 'catalog'.

See the model registry guide. Apache-2.0.