open-multi-agent/src/llm/minimax.ts

479 lines
15 KiB
TypeScript

/**
* @fileoverview MiniMax adapter implementing {@link LLMAdapter}.
*
* MiniMax provides an OpenAI-compatible Chat Completions API at
* `https://api.minimax.io/v1`, so this adapter delegates to the `openai` SDK
* with a custom `baseURL` and handles MiniMax-specific constraints:
*
* - **Temperature** must be in the open interval (0, 1]. A caller-supplied
* value of `0` is clamped to `0.01` (deterministic-ish) and values above
* `1` are clamped to `1`. When temperature is omitted the API default
* applies (the SDK omits the field rather than sending `undefined`).
*
* API key resolution order:
* 1. `apiKey` constructor argument
* 2. `MINIMAX_API_KEY` environment variable
*
* Supported models (204 K context window):
* - `MiniMax-M2.7` — latest, highest capability
* - `MiniMax-M2.7-highspeed` — faster, lower latency
* - `MiniMax-M2.5` — previous generation
* - `MiniMax-M2.5-highspeed` — previous generation, faster
*
* @example
* ```ts
* import { MiniMaxAdapter } from './minimax.js'
*
* const adapter = new MiniMaxAdapter()
* const response = await adapter.chat(messages, {
* model: 'MiniMax-M2.7',
* maxTokens: 2048,
* })
* ```
*/
import OpenAI from 'openai'
import type {
ChatCompletion,
ChatCompletionAssistantMessageParam,
ChatCompletionChunk,
ChatCompletionMessageParam,
ChatCompletionMessageToolCall,
ChatCompletionTool,
ChatCompletionToolMessageParam,
ChatCompletionUserMessageParam,
} from 'openai/resources/chat/completions/index.js'
import type {
ContentBlock,
LLMAdapter,
LLMChatOptions,
LLMMessage,
LLMResponse,
LLMStreamOptions,
LLMToolDef,
StreamEvent,
TextBlock,
ToolUseBlock,
} from '../types.js'
// ---------------------------------------------------------------------------
// Constants
// ---------------------------------------------------------------------------
/** Base URL for the MiniMax OpenAI-compatible Chat Completions API. */
const MINIMAX_BASE_URL = 'https://api.minimax.io/v1'
/**
* MiniMax requires temperature in the open interval (0, 1].
* Clamp zero → this floor and values above 1 → 1.
*/
const TEMP_FLOOR = 0.01
const TEMP_CEIL = 1.0
// ---------------------------------------------------------------------------
// Temperature helper
// ---------------------------------------------------------------------------
/**
* Clamp a temperature value to MiniMax's accepted range (0, 1].
*
* - `undefined` → `undefined` (let the API use its default)
* - `<= 0` → {@link TEMP_FLOOR} (0.01)
* - `> 1` → {@link TEMP_CEIL} (1.0)
* - otherwise → unchanged
*/
function clampTemperature(temperature: number | undefined): number | undefined {
if (temperature === undefined) return undefined
if (temperature <= 0) return TEMP_FLOOR
if (temperature > TEMP_CEIL) return TEMP_CEIL
return temperature
}
// ---------------------------------------------------------------------------
// Internal helpers — framework → OpenAI wire format
// ---------------------------------------------------------------------------
function toMiniMaxTool(tool: LLMToolDef): ChatCompletionTool {
return {
type: 'function',
function: {
name: tool.name,
description: tool.description,
parameters: tool.inputSchema as Record<string, unknown>,
},
}
}
function hasToolResults(msg: LLMMessage): boolean {
return msg.content.some((b) => b.type === 'tool_result')
}
function toMiniMaxMessages(messages: LLMMessage[]): ChatCompletionMessageParam[] {
const result: ChatCompletionMessageParam[] = []
for (const msg of messages) {
if (msg.role === 'assistant') {
result.push(toMiniMaxAssistantMessage(msg))
} else {
if (!hasToolResults(msg)) {
result.push(toMiniMaxUserMessage(msg))
} else {
const nonToolBlocks = msg.content.filter((b) => b.type !== 'tool_result')
if (nonToolBlocks.length > 0) {
result.push(toMiniMaxUserMessage({ role: 'user', content: nonToolBlocks }))
}
for (const block of msg.content) {
if (block.type === 'tool_result') {
const toolMsg: ChatCompletionToolMessageParam = {
role: 'tool',
tool_call_id: block.tool_use_id,
content: block.content,
}
result.push(toolMsg)
}
}
}
}
}
return result
}
function toMiniMaxUserMessage(msg: LLMMessage): ChatCompletionUserMessageParam {
if (msg.content.length === 1 && msg.content[0]?.type === 'text') {
return { role: 'user', content: msg.content[0].text }
}
type ContentPart = OpenAI.Chat.ChatCompletionContentPartText | OpenAI.Chat.ChatCompletionContentPartImage
const parts: ContentPart[] = []
for (const block of msg.content) {
if (block.type === 'text') {
parts.push({ type: 'text', text: block.text })
} else if (block.type === 'image') {
parts.push({
type: 'image_url',
image_url: {
url: `data:${block.source.media_type};base64,${block.source.data}`,
},
})
}
}
return { role: 'user', content: parts }
}
function toMiniMaxAssistantMessage(msg: LLMMessage): ChatCompletionAssistantMessageParam {
const toolCalls: ChatCompletionMessageToolCall[] = []
const textParts: string[] = []
for (const block of msg.content) {
if (block.type === 'tool_use') {
toolCalls.push({
id: block.id,
type: 'function',
function: {
name: block.name,
arguments: JSON.stringify(block.input),
},
})
} else if (block.type === 'text') {
textParts.push(block.text)
}
}
const assistantMsg: ChatCompletionAssistantMessageParam = {
role: 'assistant',
content: textParts.length > 0 ? textParts.join('') : null,
}
if (toolCalls.length > 0) {
assistantMsg.tool_calls = toolCalls
}
return assistantMsg
}
// ---------------------------------------------------------------------------
// Internal helpers — OpenAI wire format → framework
// ---------------------------------------------------------------------------
function fromMiniMaxCompletion(completion: ChatCompletion): LLMResponse {
const choice = completion.choices[0]
if (choice === undefined) {
throw new Error('MiniMax returned a completion with no choices')
}
const content: ContentBlock[] = []
const message = choice.message
if (message.content !== null && message.content !== undefined) {
const textBlock: TextBlock = { type: 'text', text: message.content }
content.push(textBlock)
}
for (const toolCall of message.tool_calls ?? []) {
let parsedInput: Record<string, unknown> = {}
try {
const parsed: unknown = JSON.parse(toolCall.function.arguments)
if (parsed !== null && typeof parsed === 'object' && !Array.isArray(parsed)) {
parsedInput = parsed as Record<string, unknown>
}
} catch {
// Malformed arguments — surface as empty object.
}
const toolUseBlock: ToolUseBlock = {
type: 'tool_use',
id: toolCall.id,
name: toolCall.function.name,
input: parsedInput,
}
content.push(toolUseBlock)
}
return {
id: completion.id,
content,
model: completion.model,
stop_reason: normalizeFinishReason(choice.finish_reason ?? 'stop'),
usage: {
input_tokens: completion.usage?.prompt_tokens ?? 0,
output_tokens: completion.usage?.completion_tokens ?? 0,
},
}
}
function normalizeFinishReason(reason: string): string {
switch (reason) {
case 'stop': return 'end_turn'
case 'tool_calls': return 'tool_use'
case 'length': return 'max_tokens'
case 'content_filter': return 'content_filter'
default: return reason
}
}
// ---------------------------------------------------------------------------
// Adapter implementation
// ---------------------------------------------------------------------------
/**
* LLM adapter backed by the MiniMax Chat Completions API.
*
* Uses the OpenAI SDK pointed at `https://api.minimax.io/v1`.
* Thread-safe — a single instance may be shared across concurrent agent runs.
*/
export class MiniMaxAdapter implements LLMAdapter {
readonly name = 'minimax'
readonly #client: OpenAI
constructor(apiKey?: string) {
this.#client = new OpenAI({
apiKey: apiKey ?? process.env['MINIMAX_API_KEY'],
baseURL: MINIMAX_BASE_URL,
})
}
// -------------------------------------------------------------------------
// chat()
// -------------------------------------------------------------------------
/**
* Send a synchronous (non-streaming) chat request and return the complete
* {@link LLMResponse}.
*
* Temperature is clamped to MiniMax's accepted range (0, 1] before sending.
*/
async chat(messages: LLMMessage[], options: LLMChatOptions): Promise<LLMResponse> {
const miniMaxMessages = buildMiniMaxMessageList(messages, options.systemPrompt)
const completion = await this.#client.chat.completions.create(
{
model: options.model,
messages: miniMaxMessages,
max_tokens: options.maxTokens,
temperature: clampTemperature(options.temperature),
tools: options.tools ? options.tools.map(toMiniMaxTool) : undefined,
stream: false,
},
{
signal: options.abortSignal,
},
)
return fromMiniMaxCompletion(completion)
}
// -------------------------------------------------------------------------
// stream()
// -------------------------------------------------------------------------
/**
* Send a streaming chat request and yield {@link StreamEvent}s incrementally.
*
* Temperature is clamped to MiniMax's accepted range (0, 1] before sending.
*
* Sequence guarantees:
* - Zero or more `text` events
* - Zero or more `tool_use` events (emitted once per tool call, after
* arguments have been fully assembled)
* - Exactly one terminal event: `done` or `error`
*/
async *stream(
messages: LLMMessage[],
options: LLMStreamOptions,
): AsyncIterable<StreamEvent> {
const miniMaxMessages = buildMiniMaxMessageList(messages, options.systemPrompt)
let completionId = ''
let completionModel = ''
let finalFinishReason: string = 'stop'
let inputTokens = 0
let outputTokens = 0
const toolCallBuffers = new Map<
number,
{ id: string; name: string; argsJson: string }
>()
let fullText = ''
try {
const streamResponse = await this.#client.chat.completions.create(
{
model: options.model,
messages: miniMaxMessages,
max_tokens: options.maxTokens,
temperature: clampTemperature(options.temperature),
tools: options.tools ? options.tools.map(toMiniMaxTool) : undefined,
stream: true,
stream_options: { include_usage: true },
},
{
signal: options.abortSignal,
},
)
for await (const chunk of streamResponse) {
completionId = chunk.id
completionModel = chunk.model
if (chunk.usage !== null && chunk.usage !== undefined) {
inputTokens = chunk.usage.prompt_tokens
outputTokens = chunk.usage.completion_tokens
}
const choice: ChatCompletionChunk.Choice | undefined = chunk.choices[0]
if (choice === undefined) continue
const delta = choice.delta
if (delta.content !== null && delta.content !== undefined) {
fullText += delta.content
yield { type: 'text', data: delta.content } satisfies StreamEvent
}
for (const toolCallDelta of delta.tool_calls ?? []) {
const idx = toolCallDelta.index
if (!toolCallBuffers.has(idx)) {
toolCallBuffers.set(idx, {
id: toolCallDelta.id ?? '',
name: toolCallDelta.function?.name ?? '',
argsJson: '',
})
}
const buf = toolCallBuffers.get(idx)
if (buf !== undefined) {
if (toolCallDelta.id) buf.id = toolCallDelta.id
if (toolCallDelta.function?.name) buf.name = toolCallDelta.function.name
if (toolCallDelta.function?.arguments) {
buf.argsJson += toolCallDelta.function.arguments
}
}
}
if (choice.finish_reason !== null && choice.finish_reason !== undefined) {
finalFinishReason = choice.finish_reason
}
}
const finalToolUseBlocks: ToolUseBlock[] = []
for (const buf of toolCallBuffers.values()) {
let parsedInput: Record<string, unknown> = {}
try {
const parsed: unknown = JSON.parse(buf.argsJson)
if (parsed !== null && typeof parsed === 'object' && !Array.isArray(parsed)) {
parsedInput = parsed as Record<string, unknown>
}
} catch {
// Malformed JSON — surface as empty object.
}
const toolUseBlock: ToolUseBlock = {
type: 'tool_use',
id: buf.id,
name: buf.name,
input: parsedInput,
}
finalToolUseBlocks.push(toolUseBlock)
yield { type: 'tool_use', data: toolUseBlock } satisfies StreamEvent
}
const doneContent: ContentBlock[] = []
if (fullText.length > 0) {
doneContent.push({ type: 'text', text: fullText } satisfies TextBlock)
}
doneContent.push(...finalToolUseBlocks)
const finalResponse: LLMResponse = {
id: completionId,
content: doneContent,
model: completionModel,
stop_reason: normalizeFinishReason(finalFinishReason),
usage: { input_tokens: inputTokens, output_tokens: outputTokens },
}
yield { type: 'done', data: finalResponse } satisfies StreamEvent
} catch (err) {
const error = err instanceof Error ? err : new Error(String(err))
yield { type: 'error', data: error } satisfies StreamEvent
}
}
}
// ---------------------------------------------------------------------------
// Private utility
// ---------------------------------------------------------------------------
function buildMiniMaxMessageList(
messages: LLMMessage[],
systemPrompt: string | undefined,
): ChatCompletionMessageParam[] {
const result: ChatCompletionMessageParam[] = []
if (systemPrompt !== undefined && systemPrompt.length > 0) {
result.push({ role: 'system', content: systemPrompt })
}
result.push(...toMiniMaxMessages(messages))
return result
}
// Re-export types that consumers of this module commonly need alongside the adapter.
export type {
ContentBlock,
LLMAdapter,
LLMChatOptions,
LLMMessage,
LLMResponse,
LLMStreamOptions,
LLMToolDef,
StreamEvent,
}