We rank for usefulness and signal, not for raw collection size.
Not the biggest directory. Just today's most useful AI agent signals.
FindAI.run helps both humans and agents answer three questions fast: what matters today, what it does, and how to start.
How to use this site
Start with today's picksThe top four items are the best signal-to-noise ratio for today.
Check what it can doEach item tries to explain the actual use case instead of only naming the thing.
Then decide whether to startWe compress the first actionable step so you can move without opening ten tabs.
Every pick should tell you the practical use case in one glance.
We compress the first step so you can act without opening ten tabs.
Today's picks
The most useful agent signals to act on right now.
Start here#docs#openclaw#operations
OpenClaw Docs
The main documentation hub for OpenClaw setup, architecture, channels, skills, and operations.
What it can do
Explains how OpenClaw is structured, configured, and extended in actual deployments.
How to start
Start with setup, channels, and skills. Then bookmark the docs for follow-up troubleshooting.
Why it matters
If you are building around OpenClaw, this is still the fastest way to get the real model right.
Useful#skills#registry#discovery
ClawHub Skill Registry
A public registry for discovering and installing agent skills when you need to extend an agent fast.
What it can do
Lets you discover installable skills by task, workflow, and capability instead of building everything from scratch.
How to start
Search by the task you want done, shortlist two or three candidates, and test the most specific one first.
Why it matters
It shortens the path from idea to usable capability, especially for personal agents.
Core#github#source#openclaw
OpenClaw GitHub Repository
The main source repo behind OpenClaw. 316256 stars, updated 2026-03-16. Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞
What it can do
Lets you track releases, issues, source changes, and the actual implementation behind the OpenClaw ecosystem.
How to start
Watch releases first, then inspect issues and the repo tree when a feature or behavior matters to you.
Why it matters
This is still the highest-signal public source if you care about what OpenClaw is actually shipping.
MCP#mcp#servers#awesome
Awesome MCP Servers
A curated list of MCP servers and related tooling. 83228 stars, updated 2026-03-16. A collection of MCP servers.
What it can do
Acts like a fast-moving map of MCP tooling, so you can spot mature integrations without crawling the whole ecosystem yourself.
How to start
Skim the list by domain, open two or three repositories that match your workflow, and test the simplest one first.
Why it matters
MCP is noisy right now, and this is one of the fastest ways to reduce search cost.
Latest feed
The newest signals across the site, sorted by freshness.
1
Core#github#source#openclaw
OpenClaw GitHub Repository
The main source repo behind OpenClaw. 316256 stars, updated 2026-03-16. Your own personal AI assistant. Any OS. Any Platform. The lobster way. 🦞
Lets you track releases, issues, source changes, and the actual implementation behind the OpenClaw ecosystem.
2
Practical#browser#automation#agent
browser-use
An agent-focused browser automation project. 80910 stars, updated 2026-03-16. 🌐 Make websites accessible for AI agents. Automate tasks online with ease.
Turns browser automation into an agent-native workflow instead of a brittle script-only setup.
3
SDK#agents#python#sdk
OpenAI Agents SDK (Python)
A practical Python SDK for building agent workflows. 20029 stars, updated 2026-03-16. A lightweight, powerful framework for multi-agent workflows
Shows how a modern agent SDK is structured in Python, including tools, handoffs, and orchestration patterns.
4
Reference#mcp#reference#servers
Model Context Protocol Servers
Reference implementations and examples for MCP servers. 81203 stars, updated 2026-03-16. Model Context Protocol Servers
Provides reference code and examples you can inspect when you want to understand how MCP servers are actually shaped.
5
Bridge#mcp#playwright#browser
Playwright MCP
A Playwright-based MCP project for browser interaction. 29003 stars, updated 2026-03-16. Playwright MCP server
Brings browser interaction into an MCP-shaped interface, which is a useful bridge between automation and agent tooling.
6
MCP#mcp#servers#awesome
Awesome MCP Servers
A curated list of MCP servers and related tooling. 83228 stars, updated 2026-03-16. A collection of MCP servers.
Acts like a fast-moving map of MCP tooling, so you can spot mature integrations without crawling the whole ecosystem yourself.
7
Start here#docs#openclaw#operations
OpenClaw Docs
The main documentation hub for OpenClaw setup, architecture, channels, skills, and operations.
Explains how OpenClaw is structured, configured, and extended in actual deployments.
8
Useful#skills#registry#discovery
ClawHub Skill Registry
A public registry for discovering and installing agent skills when you need to extend an agent fast.
Lets you discover installable skills by task, workflow, and capability instead of building everything from scratch.
Browse by use case
If you came to solve a problem instead of browsing a directory, start here.
Extend an agent
Find skills, runtimes, and tool patterns that can be plugged into a live agent workflow.
Ship automation
Focus on practical workflows you can actually deploy, not concept demos only.
Discover MCP tools
Track MCP servers and references that are mature enough to use today.
Find fun experiments
Keep a lane for weird, fun, or social agent experiments worth trying.
Browse by content type
The site is organized around OpenClaw, Skills, MCP, and Workflows.
OpenClaw
Docs, releases, repo signals, and core updates around OpenClaw.
4 items
Skills
Installable capabilities and skill-shaped building blocks for personal agents.
2 items
MCP
MCP servers, references, and maps that reduce integration search cost.
3 items
Workflows
Interesting or practical agent workflows worth trying in the real world.
3 items
For agents
Pages for humans, endpoints for agents.
/api/latest.json
/api/featured.json
/api/categories.json
/api/openclaw-recommended.json
/api/search-index.json
/rss/en.xml
/rss/zh.xml
/llms.txt
Why agents can consume this site too
Every item is split into what it is, what it can do, and how to start, so the page is more useful than a naked link list.
The same content is also exposed as JSON, RSS, and llms.txt for OpenClaw agents and other automations.