AgenticBook
04 · Category

Agent Tools, Perception & Capabilities

Connectors, scrapers, and APIs that extend what agents can perceive and do.

68 resources4 sections
Filter

Browser & Web Automation

11
steel.dev

Open-source browser infrastructure for AI agents, providing scalable browser sessions with built-in proxies and captcha solving. Acts as a managed Puppeteer/Playwright cluster with session persistence. Self-hostable on any cloud.

Details
screenshotone.com

Screenshot API for agents that need visual context of web pages. Renders any URL to PNG or PDF with custom viewport sizes, full-page capture, and element blocking. Useful for agents that reason about page layouts.

Details
lightpanda.io

Ultra-fast headless browser written in Zig, designed for AI and automation workloads. 10x faster than Chrome with 90% less memory usage. Supports JavaScript execution and is purpose-built for agents that need to scrape thousands of pages per minute.

Details
firecrawl.dev

Web scraping API that converts any URL into clean markdown optimized for LLM consumption. Handles JavaScript rendering, authentication, and crawling entire websites recursively. The go-to data ingestion layer for RAG pipelines and research agents.

Details
hyperbrowser.ai

Scalable cloud browser infrastructure with built-in CAPTCHA solving, proxy rotation, and session management. Provides a clean REST API and MCP integration for agents that need reliable web access at scale. Handles anti-bot measures automatically.

Details
playwright.dev

Microsoft's cross-browser automation library supporting Chromium, Firefox, and WebKit. Provides a high-level API for reliable web scraping, testing, and browser automation. Built-in network interception, trace viewer, and codegen for generating automation scripts.

Details
browserbase.com

Managed headless browser infrastructure for AI agents. Handles browser lifecycle, session management, and anti-bot challenges. Provides a Playwright/Puppeteer-compatible API without managing browser infrastructure. Includes session recording, debugging, and Live View.

Details
agentql.com

Query language for AI agents to interact with web pages semantically. Define what data or elements you want in natural language; AgentQL finds them even as the page changes. Works with Playwright and Puppeteer to make browser automation more resilient.

Details
github.com

Official Microsoft MCP server exposing Playwright browser automation to AI agents. Enables agents using Claude, GPT, and other MCP-compatible clients to navigate the web, fill forms, and extract data using accessibility-tree snapshots for efficiency.

Details
github.com

Alibaba's Qwen-powered web browsing agent. Combines web search, page reading, and multi-step reasoning to answer complex questions requiring information from multiple sources. Demonstrates how modern LLMs can use browser tools for extended research tasks.

Details
crawlee.dev

Web scraping and browser automation library from Apify. Provides request queues, storage, proxy management, and anti-blocking features out of the box. Supports Playwright, Puppeteer, and Cheerio crawlers. Ideal for large-scale web data collection for agent knowledge bases.

Details

Advanced Scraping & LLM-Ready Web Data

12
apify.com

Web scraping and automation platform with a marketplace of 1,500+ pre-built scrapers (Actors). Provides cloud infrastructure for running crawlers at scale with proxy management, scheduling, and result storage. REST API makes it easy to integrate scraped data into agent pipelines.

Details
scrapingbee.com

Web scraping API handling JavaScript rendering, CAPTCHA solving, and proxy rotation. Simple REST API supporting headless Chrome for dynamic content extraction. AI extraction endpoint uses LLMs to extract structured data from unstructured web pages.

Details
spider.cloud

Fastest web crawler purpose-built for LLM data pipelines. Outputs clean, structured markdown with lightning-fast parallel crawling across thousands of pages — optimized for throughput rather than stealth, making it ideal for large-scale knowledge base construction.

Details
brightdata.com

Enterprise web data platform with the world's largest proxy network (72M+ IPs). Provides structured datasets, scraping browser, and web scraper IDE. Compliance-focused with ethical data collection policies. Used by enterprises for large-scale web data acquisition for agent training and grounding.

Details
zenrows.com

Web scraping API with anti-bot bypass, headless browser rendering, and CSS selector/XPath extraction. Returns clean HTML or JSON. Premium proxies with automatic rotation. Competitive pricing with per-request billing and a generous free tier for development.

Details
mendable.ai

AI search and chat platform backed by your indexed documentation and knowledge bases. Provides a managed search layer that agents can query to retrieve relevant documentation snippets — used by major dev tools to power their AI assistant features.

Details
scrapingant.com

Headless browser scraping API using real Chrome with residential proxies. JavaScript rendering, cookie handling, and custom headers support. Asynchronous crawling API for processing multiple URLs in parallel. Simple pricing and a free tier suitable for agent prototyping.

Details
firecrawl.dev

Web scraping API that converts entire websites into clean Markdown optimized for LLM ingestion. Handles JavaScript rendering, authentication, rate limiting, and anti-bot measures. Provides crawl, scrape, search, extract, and map endpoints. Used by thousands of RAG pipelines.

Details
spider.cloud

Fastest web crawler and scraper with a focus on speed and simplicity. Returns content in Markdown, HTML, or JSON formats. Supports crawling entire sites, extracting links, and AI-powered extraction. Competitive pricing with a free tier and simple REST API.

Details
oxylabs.io

Enterprise proxy and web scraping solutions provider. Residential, datacenter, and mobile proxies with 100M+ IP pool. AI-powered Web Scraper API with automatic parsing. SERP scraper for search engine data collection. Strong compliance and legal framework for enterprise use.

Details

Free API (r.jina.ai) that converts any URL to clean, LLM-friendly text by prepending r.jina.ai/ to the URL. No API key required for basic use. Handles complex pages, PDFs, and extracts meaningful content stripping away navigation and ads.

Details
github.com

Open-source web crawler designed specifically for AI data collection. Produces clean, LLM-ready output from websites with intelligent content extraction, session-based crawling, and structured data extraction using LLMs. Docker-ready with a REST API and Python library.

Details

Unified Tool & API Connectors

29
toolhouse.ai

Cloud infrastructure for giving LLMs the ability to use tools with a universal tool store. Provides pre-built tools for web search, code execution, email, and more as MCP or function-call compatible APIs. Handles auth, caching, and rate limiting.

Details
airbyte.com

Open-source data integration platform for moving data from 300+ sources into warehouses and vector databases. Essential for building RAG pipelines that need fresh, synced data from CRMs, databases, and SaaS tools. Used at scale by thousands of companies.

Details
mcpservers.org

Community registry of Model Context Protocol servers, providing agents with access to hundreds of tools and data sources. Includes servers for filesystems, databases, APIs, and cloud services. The npm registry for the MCP ecosystem.

Details

Visual workflow automation platform with 1800+ app integrations. Connects AI agent outputs to downstream business tools through a drag-and-drop scenario builder. Supports complex branching logic and data transformation.

Details
unify.ai

LLM routing and benchmarking platform that automatically routes queries to the best model for quality, cost, and speed. Provides a unified API across 40+ LLM providers with per-query performance analytics. Reduces LLM costs with intelligent routing.

Details
merge.dev

Unified API providing a single integration point for 180+ HR, ATS, CRM, accounting, and ticketing systems. AI agents use Merge to read and write data across enterprise tools without handling individual API integrations. Enterprise-grade with data normalization.

Details
morph-data.io

Framework for building data apps and agent tools on top of SQL databases and Python functions. Exposes database queries and transformations as typed tools for AI agents. Bridges the gap between structured enterprise data and agentic workflows.

Details
browserbase.com

Cloud browser platform with an official MCP server giving agents persistent, stealth browser sessions with built-in proxy rotation. Maintains session cookies across multiple agent turns for stateful web interaction. Used by leading AI coding tools.

Details

Stripe's official SDK for integrating payment capabilities into AI agents. Provides tools for creating customers, charges, invoices, and subscriptions as LLM function calls. Works with OpenAI, Anthropic, Vercel AI SDK, and LangChain out of the box.

Details

GitHub's official MCP server providing agents with tools for reading repos, creating issues, opening PRs, and searching code. Enables coding agents to interact with GitHub natively through the MCP protocol. The most widely used MCP server.

Details
github.com

Microsoft's official MCP server exposing Playwright browser automation as tools for AI agents. Enables agents to navigate pages, click, fill forms, and extract content using structured accessibility snapshots. Zero-setup MCP integration for browser-capable agents.

Details
runalloy.com

Embedded integration platform for adding 200+ e-commerce, marketing, and logistics integrations to AI agents. Provides a white-label UI for end-users to connect their own accounts. Popular with AI agents serving e-commerce workflows.

Details
composio.dev

Managed tool platform providing 250+ authenticated integrations (GitHub, Slack, Gmail, Jira, Salesforce) as ready-to-use tools for AI agents. Handles OAuth flows, credential management, and keeps integrations up to date. Native support for LangChain, CrewAI, AutoGen, and Claude.

Details
github.com

Open-source training dataset and execution environment for LLM tool use. Contains 16,000+ real-world APIs across 49 categories with instruction-following data — used to fine-tune models that can accurately select and call the right tool for any user task.

Details
nango.dev

Open-source unified API platform handling OAuth, token refresh, webhooks, and bi-directional sync for 250+ APIs. Build integrations with external services in hours, not weeks. Syncs data to your database in real-time. Self-hostable or managed cloud.

Details

Expose 6,000+ Zapier-supported apps as callable actions for any AI agent. Lets agents trigger workflows across Gmail, Google Sheets, Notion, Airtable, and hundreds of other business tools through a natural language interface — no custom integration code needed.

Details
zapier.com

Access 6,000+ Zapier-connected apps from AI agents via the Model Context Protocol. Define allowed actions and connect agents to CRMs, email, databases, and productivity tools without writing integration code. Works with Claude, GPT, and any MCP-compatible agent.

Details
gorilla.cs.berkeley.edu

LLM fine-tuned specifically to invoke 1,600+ APIs with high accuracy. Achieves state-of-the-art performance on API selection and parameter filling benchmarks — a research project from Berkeley showing how targeted fine-tuning improves tool-use reliability.

Details

Visual no-code integration platform connecting 1,500+ apps with complex logic support. HTTP/webhook modules enable AI agents to trigger and receive automation flows. Ideal for connecting agents to business workflows without custom API integration code.

Details
python.langchain.com

Standardized tool interface and extensive catalog of pre-built integrations for LangChain agents. Covers search engines, code execution, file I/O, databases, and third-party APIs — with a simple decorator-based API for wrapping any function as an agent-callable tool.

Details
n8n.io

Open-source workflow automation platform with built-in AI nodes (LangChain, embeddings, vector stores). Self-hostable alternative to Zapier/Make with full source control. AI Agent node enables building agentic workflows that respond to triggers and use 400+ integrations.

Details
modelcontextprotocol.io

Open standard by Anthropic for connecting LLMs to external tools, data sources, and APIs. Servers expose tools, resources, and prompts; clients (Claude, IDE plugins, agent frameworks) connect to them. Rapidly becoming the universal plugin standard for AI agent tooling.

Details
activepieces.com

Open-source Zapier alternative with 200+ integrations and a clean visual builder. Business automation platform increasingly used to create trigger-action workflows for AI agents. TypeScript SDK for building custom pieces (integrations) for proprietary systems.

Details
pipedream.com

Developer-first integration platform combining no-code triggers with custom Node.js/Python code steps. 2,500+ app integrations. Deploy server-side code without managing infrastructure. Popular for connecting AI agent outputs to business systems via webhooks and scheduled triggers.

Details
segment.com

Customer data platform that provides a unified API to collect, clean, and route user events to 300+ analytics, marketing, and data tools. Agent applications can leverage Segment to track user interactions and personalize behavior based on customer data.

Details
composio.dev

Integration platform for AI Agents and LLMs with 200+ tool connectors. Lets agents access GitHub, Slack, Gmail, databases, and more through a unified auth layer. Free tier available.

Details
browsercat.com

Headless browser API for automation, scraping, and AI agent web access. Supports image and PDF generation. Free plan includes 1k requests/month.

Details
tavily.com

Search API purpose-built for LLMs and AI agents, returning clean structured results optimised for RAG pipelines. 1,000 requests/month free, no credit card required.

Details
getmaxim.ai

End-to-end evaluation and observability platform for AI agents. Simulate, evaluate, and monitor agents reliably. Free forever for indie developers and small teams up to 3 seats.

Details