Assistants API
Supercompat also supports the older Assistants API surface —
beta.assistants, beta.threads, and the run loop. Useful when you have an existing app built on that API and want to change the backend, or when you want stateful threads with minimal app-side bookkeeping.Set up a client
import OpenAI from 'openai'
import { PrismaClient } from '@prisma/client'
import {
supercompat,
openaiClientAdapter,
completionsRunAdapter,
prismaStorageAdapter,
} from 'supercompat/openai'
const client = supercompat({
clientAdapter: openaiClientAdapter({ openai: new OpenAI() }),
storageAdapter: prismaStorageAdapter({ prisma: new PrismaClient() }),
runAdapter: completionsRunAdapter(),
})
completionsRunAdapter makes the run loop execute against /chat/completions. That means it works against every provider — including OpenAI — because it never depends on the provider having a native assistants/runs surface.Create an assistant
const assistant = await client.beta.assistants.create({
model: 'gpt-4.1-mini',
name: 'Reviewer',
instructions: 'You review short pieces of text for clarity.',
})
Send a message and run
const thread = await client.beta.threads.create()
await client.beta.threads.messages.create(thread.id, {
role: 'user',
content: 'Please review: "The quick brown fox."',
})
const run = await client.beta.threads.runs.createAndPoll(thread.id, {
assistant_id: assistant.id,
})
const messages = await client.beta.threads.messages.list(thread.id)
console.log(messages.data[0].content)
createAndPoll waits for the run to complete. Under the hood, Supercompat is making chat completion calls through your chosen provider and writing thread/run/message state through the storage adapter.Against a different provider
import Anthropic from '@anthropic-ai/sdk'
import { anthropicClientAdapter } from 'supercompat/openai'
const client = supercompat({
clientAdapter: anthropicClientAdapter({ anthropic: new Anthropic() }),
storageAdapter: prismaStorageAdapter({ prisma: new PrismaClient() }),
runAdapter: completionsRunAdapter(),
})
// Everything below works against Claude.
const assistant = await client.beta.assistants.create({
model: 'claude-sonnet-4-6',
name: 'Reviewer',
instructions: 'You review short pieces of text for clarity.',
})
Streaming a run
const stream = await client.beta.threads.runs.create(thread.id, {
assistant_id: assistant.id,
stream: true,
})
for await (const event of stream) {
if (event.event === 'thread.message.delta') {
const delta = event.data.delta.content?.[0]
if (delta?.type === 'text') process.stdout.write(delta.text.value)
}
}
Events are
OpenAI.Beta.AssistantStreamEvent on every provider.Next
Tools — attach function tools to an assistant or a run.
Storage adapters — compare Prisma, OpenAI-managed, and Azure-managed state.