Xsai SDK

What it does

When your agent uses it

• Data & Analytics work in your project

Example prompts

• “/xsai-sdk”

What it can do on your machine

Read from SKILL.md and the folder at commit 3d2834d. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  No scripts in the folder and no shell commands in SKILL.md (its code samples are typescript).

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  Hosts in commands or code, which the agent is likely to contact:

  • api.openai.com

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Xsai SDK loads about 1.1k tokens when it runs, and up to ~44k if it reads all its reference files. Until then it costs about 22 tokens; SKILL.md has 134 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~22

When it runs · the whole SKILL.md, loaded when a task matches

~1.1k

With references · SKILL.md plus every file in references/, read only if the agent opens them

~44k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from yusixian/moe-copy-ai at commit 3d2834d, republished under its AGPL-3.0 licence (© yusixian). 134 words, ~1,138 tokens.

name

xsai-sdk

description

xsAI SDK v0.4.x API reference - extra-small AI SDK with OpenAI-compatible API

globs

utils/ai-service.ts, hooks/useStreamProcessor.ts, hooks/useAiSummary.ts, hooks/useAiSettings.ts

import { streamText } from "@xsai/stream-text"

// NOTE: streamText is SYNCHRONOUS (not async) in 0.4.x
const { textStream } = streamText({
  apiKey: "...",
  baseURL: "https://api.openai.com/v1/",
  model: "gpt-4o",
  messages: [
    { role: "system", content: "You are a helpful assistant." },
    { role: "user", content: "Hello" },
  ],
  streamOptions: { includeUsage: true },
})

// textStream is a ReadableStream<string>, supports async iteration
for await (const textPart of textStream) {
  console.log(textPart)
}

interface StreamTextResult {
  textStream: ReadableStream<string>
  fullStream: ReadableStream<StreamTextEvent>
  reasoningTextStream: ReadableStream<string>
  messages: Promise<Message[]>
  steps: Promise<CompletionStep[]>
  usage: Promise<undefined | Usage>
  totalUsage: Promise<undefined | Usage>
}

interface StreamTextOptions extends ChatOptions {
  maxSteps?: number  // default: 1
  onEvent?: (event: StreamTextEvent) => Promise<unknown> | unknown
  onFinish?: (step?: CompletionStep) => Promise<unknown> | unknown
  onStepFinish?: (step: CompletionStep) => Promise<unknown> | unknown
  streamOptions?: { includeUsage?: boolean }
}

// StreamTextEvent replaces the old StreamTextChunkResult
type StreamTextEvent = /* union of stream event types */

import { generateText } from "@xsai/generate-text"

const { text, usage } = await generateText({
  apiKey: "...",
  baseURL: "https://api.openai.com/v1/",
  model: "gpt-4o",
  messages: [
    { role: "system", content: "You are a helpful assistant." },
    { role: "user", content: "Why is the sky blue?" },
  ],
})

interface GenerateTextResult {
  text?: string  // NOTE: optional in 0.4.x (was required in 0.2.x)
  usage: Usage
  finishReason: FinishReason
  messages: Message[]
  steps: CompletionStep<true>[]
  toolCalls: CompletionToolCall[]
  toolResults: CompletionToolResult[]
  reasoningText?: string
}

import { listModels } from "@xsai/model"

const models = await listModels({
  apiKey: "...",
  baseURL: "https://api.openai.com/v1/",
})
// Returns Model[] with { id, object, created, owned_by }

// From @xsai/shared-chat
interface ChatOptions {
  apiKey?: string
  baseURL: string
  model: string
  messages: Message[]
  // ... other OpenAI-compatible options
}

interface Usage {
  prompt_tokens: number
  completion_tokens: number
  total_tokens: number
}

type Message = SystemMessage | UserMessage | AssistantMessage | ToolMessage | DeveloperMessage

import { smoothStream } from "@xsai/utils-stream"

const smoothed = textStream.pipeThrough(smoothStream({
  delay: 20,
  chunking: "line",
}))

import { toAsyncIterator } from "@xsai/utils-stream"

// For browsers that don't support ReadableStream async iteration
const iterableStream = toAsyncIterator(textStream)

© yusixian, AGPL-3.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

SKILL.md and 48 other files (references) in .claude/skills/xsai of yusixian/moe-copy-ai.

• SKILL.md
• references/docs/Chat.md
• references/docs/Composio.md
• references/docs/Embeddings.md
• references/docs/Image.md
• references/docs/Model-Context-Protocol.md
• references/docs/Models.md
• references/docs/Overview.md
• references/docs/Providers.md
• references/docs/Quick-Start.md
• references/docs/Reasoning.md
• references/docs/Speech.md
• references/docs/Stream.md
• references/docs/Structured-Data-1.md
• references/docs/Structured-Data.md
• references/docs/Telemetry.md
• references/docs/Text-1.md
• references/docs/Text.md
• references/docs/Tool-Calling.md
• … and 30 more

Open the folder on GitHubat commit 3d2834d