Skip to main content

uploadSkill

Uploads a skill (a bundle of files) to a provider and returns a result whose providerReference (a ProviderReference) can be used in subsequent inference calls.

import { uploadSkill } from 'ai';
import { anthropic } from '@ai-sdk/anthropic';
import { readFileSync } from 'fs';

const { providerReference } = await uploadSkill({
  api: anthropic.skills(),
  files: [
    {
      path: 'my-skill/SKILL.md',
      data: readFileSync('./SKILL.md'),
    },
  ],
  displayTitle: 'My Skill',
});

Import

import { uploadSkill } from "ai"

API Signature

Parameters

  • api (SkillsV4 | ProviderV4): The skills API interface to use for uploading. Can be a SkillsV4 instance (e.g. anthropic.skills()) or a provider instance directly (e.g. anthropic), in which case .skills() is called automatically.
  • files (UploadSkillFile[]): The files that make up the skill. Each file must have a relative path and data. The data can be a string, a Uint8Array, or an object with type: 'text' and a string text value, or type: 'data' and a string data value.
  • displayTitle? (string): Human-readable title for the skill.
  • providerOptions? (ProviderOptions): Additional provider-specific options.

Returns

  • providerReference (ProviderReference): A Record<string, string> mapping provider names to provider-specific skill identifiers. Pass this when referencing the skill in inference calls.
  • displayTitle? (string): Human-readable title returned by the provider (if supported).
  • name? (string): Skill name inferred by the provider from the uploaded files.
  • description? (string): Skill description inferred by the provider from the uploaded files.
  • latestVersion? (string): Latest version identifier assigned by the provider.
  • providerMetadata? (ProviderMetadata): Additional provider-specific metadata returned from the upload.
  • warnings (Warning[]): Warnings from the provider (e.g. unsupported settings).
Edit on GitHub