Anvia

Media input ​

A completion model can inspect images and file documents when its adapter and model capabilities support those content types.

1. Send an image URL ​

ts
import type { UserMessage } from '@anvia/core'

const prompt: UserMessage = {
  role: 'user',
  content: [
    { type: 'text', text: 'Describe the chart and identify any sudden conversion change.' },
    {
      type: 'image',
      image: { type: 'url', url: signedImageUrl },
      detail: 'high',
    },
  ],
}

const result = await visionAgent.generate({ messages: [prompt] })

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

The provider fetches a URL source, so it must remain reachable for the request. Prefer a short-lived signed URL over a permanently public asset.

Image detail accepts auto, low, or high. Exact support remains provider- and model-specific.

2. Send bounded in-memory data ​

ts
import { Buffer } from 'node:buffer'
import type { UserMessage } from '@anvia/core'

const imageBase64 = Buffer
  .from(upload.bytes)
  .toString('base64')

const prompt: UserMessage = {
  role: 'user',
  content: [
    { type: 'text', text: 'Read the serial number in this photo.' },
    {
      type: 'image',
      image: { type: 'data', data: imageBase64 },
      mediaType: upload.mediaType,
      detail: 'high',
    },
  ],
}

const result = await visionAgent.generate({ messages: [prompt] })

Pass raw base64 without a data: URL prefix. Validate ownership, size, and detected media type before encoding. Base64 increases request size, so keep this path bounded.

3. Add a document ​

ts
const prompt: UserMessage = {
  role: 'user',
  content: [
    { type: 'text', text: 'Summarize the attached incident report.' },
    {
      type: 'file',
      data: { type: 'url', url: signedReportUrl },
      mediaType: 'application/pdf',
      filename: 'incident-report.pdf',
    },
  ],
}

const result = await documentAgent.generate({ messages: [prompt] })

Use { type: 'data', data: base64 } for bounded in-memory files and a text part for already extracted text.

For large or repeated documents, extract and chunk text or run OCR once. Sending the same file on every turn can be slower, more expensive, and harder to cite.

4. Expect capability errors ​

Anvia checks the completion model's imageInput and documentInput capability flags before sending the request. An unsupported request throws CompletionCapabilityError.

Capability support does not establish authenticity or extraction accuracy. Retain the source asset ID, ask for uncertainty where useful, and verify critical values before product writes.

Next, generate images.

Built for Anvia.