Get started

Install the provider with Core:

pnpm add @anvia/core @anvia/anthropic

Create the client on the server and pass its completion model into an agent:

import { Agent } from '@anvia/core'
import { AnthropicClient } from '@anvia/anthropic'

const anthropic = new AnthropicClient({
  apiKey: process.env.ANTHROPIC_API_KEY!,
})

const agent = new Agent({
  id: 'analyst',
  model: anthropic.completionModel({
      modelId: 'claude-sonnet-5'
  }),
  instructions: 'Analyze the evidence before answering.',
})

const result = await agent.generate({
    prompt: 'Summarize the incident report.'
})

if (result.type === 'response') {
  console.log(result.output)
}

The returned streaming completion handle works with direct completion, streaming, tools, agents, extractors, and pipeline stages.

List Anthropic models

const models = await anthropic.listModels()

Use listing for inventory and diagnostics. Keep a separate production allowlist rather than enabling every returned model automatically.

Use Vertex AI instead

import { AnthropicVertexClient } from '@anvia/anthropic'

const vertex = new AnthropicVertexClient({
  projectId: 'my-gcp-project',
  region: 'global',
})

const model = vertex.completionModel({
    modelId: 'claude-sonnet-5'
})

Vertex uses the same normalized completion adapter but a different official SDK and authentication path. It does not expose listModels().

Before production

• Keep Anthropic and Google credentials out of browser code.
• Set explicit model IDs and output limits.
• Authorize every tool inside the application, not only through its schema.
• Bound agent turns and cancellation.
• Test image input and tool behavior against the selected model.
• Review Vertex AI or compatible endpoints when not using Anthropic’s standard API.