AI SDK

AI SDK implementation for @micdrop/server.

This package provides an AI agent implementation using the Vercel AI SDK, allowing you to use any compatible language model provider.

Installation

Install AI SDK:

npm install @micdrop/ai-sdk

And install the provider you want to use, for example OpenAI:

npm install @ai-sdk/openai

AI SDK Agent

Usage with MicdropServer

import { AiSdkAgent } from '@micdrop/ai-sdk'
import { MicdropServer } from '@micdrop/server'
import { openai } from '@ai-sdk/openai' // or any other provider

const agent = new AiSdkAgent({
  model: openai('gpt-4o'), // Use any AI SDK compatible model
  systemPrompt: 'You are a helpful assistant',

  // Custom AI SDK settings (optional)
  settings: {
    temperature: 0.7,
    maxOutputTokens: 150,
  },

  // Provider specific parameters (optional)
  providerOptions: {
    openai: { reasoningEffort: 'none' },
  },
})

// Use with MicdropServer
new MicdropServer(socket, {
  agent,
  // ... other options
})

Usage without MicdropServer

import { AiSdkAgent } from '@micdrop/ai-sdk'
import { openai } from '@ai-sdk/openai'

const agent = new AiSdkAgent({
  model: openai('gpt-4o'),
  systemPrompt: 'You are a helpful assistant',
})

agent.on('Message', (message) => console.log('Message:', message))

agent.addUserMessage('Hello, what can you do?')

// The answer is a text stream, written as the model generates it
agent.answer().on('data', (chunk) => process.stdout.write(chunk))

Events

Event	Payload	Description
`Message`	`MicdropConversationItem`	A message, a tool call or a tool result was added to the conversation.
`ToolCall`	`MicdropToolCall`	A tool declared with `emitOutput` ran, with its parameters and output.
`CancelLastUserMessage`	none	The last user message was dropped because it carried no intent.
`SkipAnswer`	none	The agent stays silent and waits for the user to finish their sentence.
`EndCall`	none	The agent decided that the call is over.
`Failed`	none	The agent gave up generating an answer after its retries.

See the Agent interface for the full contract.

Supported Providers

The AI SDK Agent supports any provider compatible with the Vercel AI SDK:

OpenAI: openai('gpt-4o'), openai('gpt-3.5-turbo')
Anthropic: anthropic('claude-3-5-sonnet-20241022')
Google: google('gemini-1.5-pro'), google('gemini-1.5-flash')
Mistral: mistral('mistral-large-latest')
And many more: See AI SDK Providers

Provider specific options

settings carries the parameters every provider understands, such as the temperature. Parameters read by a single provider go in providerOptions, under the name of that provider. Turning reasoning off is the common case in a call, since reasoning delays the first token:

const agent = new AiSdkAgent({
  model: openai.chat('gpt-5.2'),
  systemPrompt: 'You are a helpful assistant',
  providerOptions: {
    openai: { reasoningEffort: 'none' },
  },
})

The same option reaches a local server behind createOpenAI. Ollama maps it to its own think flag, and a model without a reasoning mode ignores it.

Options

Option	Type	Default	Description
`model`	`LanguageModel`	Required	Any AI SDK compatible language model
`systemPrompt`	`string`	Required	System prompt for the agent
`maxRetry`	`number`	`3`	Maximum number of retries on API failures
`autoEndCall`	`boolean \| string`	`false`	Auto-detect when user wants to end call
`autoSemanticTurn`	`boolean \| string`	`false`	Handle incomplete user sentences
`autoIgnoreUserNoise`	`boolean \| string`	`false`	Filter meaningless user sounds
`extract`	`ExtractJsonOptions` \| `ExtractTagOptions`	`undefined`	Extract structured data from responses
`onBeforeAnswer`	`function`	`undefined`	Hook called before answer generation - return `true` to skip generation
`settings`	`CallSettings`	`{}`	Additional AI SDK parameters
`providerOptions`	`ProviderOptions`	`{}`	Provider specific parameters, keyed by provider name

The AI SDK Agent supports adding and removing custom tools to extend its capabilities. For detailed information about tool management, see the Tools documentation.

Advanced Features

The AI SDK Agent supports advanced features for improved conversation handling:

Auto End Call: Automatically detect when users want to end the conversation
Semantic Turn Detection: Handle incomplete sentences for natural flow
User Noise Filtering: Filter out meaningless sounds and filler words
Extract Value from Answer: Extract structured data from responses
Tools: Add custom tools to the agent