Agentic - AI Agent Standard Library

Project Overview

Agentic is a standard library of AI functions/tools optimized for both regular TypeScript usage and LLM-based usage. Agentic is compatible with all major TypeScript AI SDKs (including Vercel AI SDK, Mastra, LangChain, LlamaIndex, OpenAI SDK, MCP, etc.).

Project Address: https://github.com/transitive-bullshit/agentic

Core Features

🎯 Dual Usage Modes

• Standard TypeScript Classes: Can be used directly as regular TS classes
• LLM Tools: Can be easily integrated into various AI SDKs as tools

🔧 Broad SDK Support

Supports all major TypeScript AI SDKs:

• Vercel AI SDK
• Mastra AI Agent Framework
• LangChain
• LlamaIndex
• OpenAI SDK
• Genkit
• Dexter
• xsAI SDK
• GenAIScript
• MCP (Model Context Protocol)

✨ Key Advantages

• ✅ All tools are fully tested in production environments
• ✅ Works across all major TS AI SDKs
• ✅ Tools are hand-coded and extremely lean
• ✅ Provides excellent manual DX and LLM DX via the @aiFunction decorator
• ✅ Uses native fetch
• ✅ Uses ky to wrap fetch for easy customization of HTTP options, rate limiting, retries, etc.
• ✅ Supports tools from any MCP server
• ✅ Can generate new Agentic tool clients from OpenAPI specifications
• ✅ 100% open source, does not sell any products

Usage Examples

Using as a Regular TypeScript Class

import { WeatherClient } from '@agentic/stdlib'

// Requires setting the WEATHER_API_KEY environment variable (available for free from weatherapi.com)
const weather = new WeatherClient()
const result = await weather.getCurrentWeather({
  q: 'San Francisco'
})
console.log(result)

Using as an LLM Tool (Vercel AI SDK Example)

// SDK-specific import
import { openai } from '@ai-sdk/openai'
import { generateText } from 'ai'
import { createAISDKTools } from '@agentic/ai-sdk'

// SDK-agnostic import
import { WeatherClient } from '@agentic/stdlib'

const weather = new WeatherClient()
const result = await generateText({
  model: openai('gpt-4o-mini'),
  // Key line: use the @agentic/ai-sdk adapter
  tools: createAISDKTools(weather),
  toolChoice: 'required',
  prompt: 'What is the weather in San Francisco?'
})
console.log(result.toolResults[0])

Mixing Multiple Tools

import { SerperClient, WikipediaClient, FirecrawlClient } from '@agentic/stdlib'
import { createAIFunction } from '@agentic/core'
import { z } from 'zod'

const googleSearch = new SerperClient()
const wikipedia = new WikipediaClient()
const firecrawl = new FirecrawlClient()

const result = await generateText({
  model: openai('gpt-4o-mini'),
  // This example uses tools from 4 different sources
  tools: createAISDKTools(
    googleSearch,
    wikipedia,
    // Select a single function from the firecrawl client's AI function set
    firecrawl.functions.pick('firecrawl_search'),
    // Create a custom AI function
    createAIFunction({
      name: 'think',
      description: `Use this tool to think about something. It will not obtain new information or change the database, but just append the thought to the log. Use it when complex reasoning or some cache memory is needed.`,
      inputSchema: z.object({
        thought: z.string().describe('A thought to think about.')
      }),
      execute: ({ thought }) => thought
    })
  ),
  prompt: 'What year did Jurassic Park the movie come out, and what else happened that year?'
})

Supported Tools and Services

The project provides 50+ practical tools covering the following categories:

🔍 Search and Information Retrieval

Tool	Package Name	Description
Brave Search	@agentic/brave-search	Brave Search API
Bing	@agentic/bing	Microsoft Bing Search API
DuckDuckGo	@agentic/duck-duck-go	DuckDuckGo Search
Google Custom Search	@agentic/google-custom-search	Google Custom Search
Serper	@agentic/serper	Serper Search API
SerpAPI	@agentic/serpapi	SerpAPI Search Service
Tavily	@agentic/tavily	Tavily Search API
Exa	@agentic/exa	Exa Search API

📚 Knowledge and Research

Tool	Package Name	Description
Wikipedia	@agentic/wikipedia	Wikipedia API
Wikidata	@agentic/wikidata	Wikidata Access
ArXiv	@agentic/arxiv	ArXiv Academic Papers
Wolfram Alpha	@agentic/wolfram-alpha	Wolfram Alpha Computational Engine
HackerNews	@agentic/hacker-news	Hacker News API

🌐 Web Scraping and Processing

Tool	Package Name	Description
Firecrawl	@agentic/firecrawl	Web Scraping and Crawling
Jina	@agentic/jina	Jina Reader
Diffbot	@diffbot	Web Structure Extraction

☁️ Cloud Services and Documents

Tool	Package Name	Description
Google Drive	@agentic/google-drive	Google Drive
Google Docs	@agentic/google-docs	Google Docs
Notion	@agentic/notion	Notion API
Airtable	@agentic/airtable	Airtable Database

🌦️ Weather and Environment

Tool	Package Name	Description
Weather	@agentic/weather	Weather API
Open Meteo	@agentic/open-meteo	Open Weather Data

💼 Business and Data

Tool	Package Name	Description
Apollo	@agentic/apollo	Apollo Sales Tool
Clearbit	@agentic/clearbit	Company Data API
Hunter	@agentic/hunter	Email Finder Tool
People Data Labs	@agentic/people-data-labs	People Data API
Polygon	@agentic/polygon	Financial Market Data

🎨 Creative and Media

Tool	Package Name	Description
Midjourney	@agentic/midjourney	AI Image Generation
YouTube	@agentic/youtube	YouTube API

📱 Social and Communication

Tool	Package Name	Description
Twitter	@agentic/twitter	Twitter API
Slack	@agentic/slack	Slack API
Twilio	@agentic/twilio	Twilio Communication
Reddit	@agentic/reddit	Reddit API

💻 Development and Tools

Tool	Package Name	Description
Calculator	@agentic/calculator	Math Calculator
E2B	@agentic/e2b	Code Execution Environment
MCP Tools	@agentic/mcp	MCP Server Support

Architecture Design

Core Concepts

• AIFunctionLike: Can be any agentic client instance, a single AIFunction selected from a client's function set, or an AI function manually created via createAIFunction
• AIFunctionSet: A collection of AI functions, providing flexible function management
• Adapter Pattern: Provides a unified interface for different AI SDKs

Extensibility

• Supports automatic generation of tool clients from OpenAPI specifications
• Supports custom AI function creation
• Supports MCP server integration
• Supports function composition and selection

Installation and Usage

# Install the core library
npm install @agentic/core

# Install the standard tool library
npm install @agentic/stdlib

# Install a specific AI SDK adapter
npm install @agentic/ai-sdk      # Vercel AI SDK
npm install @agentic/langchain   # LangChain
npm install @agentic/llamaindex  # LlamaIndex

Summary

Agentic is a powerful and well-designed standard library of AI tools, providing TypeScript developers with a complete toolchain for building AI applications. Its design philosophy is "write once, run anywhere," allowing developers to seamlessly switch between different AI frameworks while providing a rich set of pre-built tools to accelerate AI application development.