@zhivex-ai/kimi

Kimi adapter for Zhivex AI SDK.

Install

bun add @zhivex-ai/kimi

Official Formula tools

Kimi official tools are exposed through Formula URIs. The SDK helpers load or declare the Formula tool schema, pass it as a Chat Completions function tool, and execute the matching Formula fiber when Kimi returns a tool call.

import { createKimi, kimiFormulaTools, kimiWebSearchTool } from "@zhivex-ai/kimi";
import { generateText } from "@zhivex-ai/core";

const kimi = createKimi({ apiKey: process.env.KIMI_API_KEY });

await generateText({
  model: kimi("kimi-k2.7-code"),
  prompt: "Search for current information and summarize it.",
  maxSteps: 2,
  tools: {
    web_search: kimiWebSearchTool()
  }
});

const tools = await kimiFormulaTools({
  apiKey: process.env.KIMI_API_KEY,
  formulas: ["moonshot/web-search:latest", "moonshot/fetch:latest"]
});

Built-in shortcuts include kimiWebSearchTool(), kimiFetchTool(), kimiCodeRunnerTool(), kimiExcelTool(), and kimiDateTool(). Current first-class model IDs are kimi-k2.7-code, kimi-k2.7-code-highspeed, kimi-k2.6, and kimi-k2.5. The older kimi-k2-0905-preview and kimi-k2-thinking IDs are kept for passthrough compatibility but are deprecated upstream.

kimi-k2.7-code and kimi-k2.7-code-highspeed always use thinking mode, so the SDK rejects reasoning.effort: "none", providerOptions.thinking.type: "disabled", custom temperature, custom top_p, and providerOptions.n values other than 1 before the request is sent. Thinking-capable Kimi models keep the official restriction that forced tool choice is unsupported while reasoning is enabled; use toolChoice: "auto" or toolChoice: "none" in that mode.

Kimi K2.7 Code and K2.6 support image and video understanding. The SDK maps shared image parts plus file parts whose media type starts with image/ or video/ into Kimi multimodal content blocks.

Repository and full documentation:

• https://github.com/Zhivex/zhivex-ai-sdk