Adapters
Adapters Overview
Adapters normalize AI provider streaming APIs into a common interface that AgentsKit hooks consume.
Adapters normalize AI provider streaming APIs into a common interface that AgentsKit hooks consume.
Built-in Adapters
import { anthropic, openai, vercelAI, generic } from '@agentskit/adapters'
// Anthropic
const adapter = anthropic({ apiKey: 'key', model: 'claude-sonnet-4-6' })
// OpenAI
const adapter = openai({ apiKey: 'key', model: 'gpt-4o' })
// Vercel AI SDK (route handler)
const adapter = vercelAI({ api: '/api/chat' })
// Generic (any ReadableStream)
const adapter = generic({
send: async (messages) => {
const res = await fetch('/api/chat', { body: JSON.stringify(messages) })
return res.body
},
})Custom Adapters
import { createAdapter } from '@agentskit/adapters'
const myAdapter = createAdapter({
send: async (messages) => fetch('/api', { body: JSON.stringify(messages) }),
parse: async function* (stream) {
const reader = stream.getReader()
// ... yield StreamChunk objects
yield { type: 'done' }
},
abort: () => { /* cleanup */ },
})From Mastra
Side-by-side migration guide. Map your Mastra code to AgentsKit — with honest callouts about where Mastra still wins.
Runtime
@agentskit/runtime is the execution engine for autonomous agents. It runs a ReAct loop — observe, think, act — until the model produces a final answer or a step limit is reached.