How we pick and rate projects

Thousands of AI projects appear on GitHub every month. Most people only hear about a few, usually through tech forums. This guide lists the ones that are clearly being used, and explains each one in plain language.

How a project gets on the list

• Every day we automatically search GitHub for AI projects that meet our criteria, including older ones that are gaining attention, and check where they're being discussed. Recently added projects carry a "New this week" label.
• Descriptions are drafted with AI from each project's own documentation, following written rules: plain words, no hype, and nothing the documentation doesn't say. Each page says whether a person has checked its summary yet.
• Projects in sensitive areas, such as face or voice copying, security testing and investing, are always checked by a person before they appear here.
• We favour projects that are actively maintained, have an open license, and that real people are using, not just bookmarking.

How the rating works

The 1–5 rating combines four public signals, refreshed daily:

• Popularity: how many people have starred the project on GitHub.
• Community: how many people have contributed to it.
• Usage: how often it's downloaded, where that number is published.
• Activity: how recently it was updated.

No single number tells the whole story, so no single signal decides the rating. When one signal is far ahead of the others, it counts for less. A rating describes public activity around a project. It isn't a judgement of its quality, safety or the people behind it.

A note on ratings

Plain-AyI is a curated collection of open-source AI projects. Ratings summarise public activity so you can compare projects at a glance. They don't rank one project as more important or more valuable than another, and being listed isn't an endorsement. A project with a lower rating may be exactly the right one for you.

With thanks

We're grateful to the developers, researchers and communities who build and maintain open-source AI, and who keep making powerful tools freely available to everyone. This site exists because of their work.

Sources

"Where people are talking about it" links to original discussions and research papers, with dates, so you can read them yourself.

Use it from your AI assistant

Step-by-step instructions with copy buttons are on the Use with AI assistants page. In short:

You can ask your AI assistant to search this guide for you, for example "find me a tool for transcribing meetings that doesn't need coding". The guide offers a free, read-only connection your assistant can use. It can look things up, but it can't change anything, and you don't need an account.

Claude app (desktop or web): open Settings, then Connectors, choose "Add custom connector", name it Plain-AyI and paste this address:

https://plain-ayi.com/mcp

Claude Code: run this in your terminal:

claude mcp add --transport http plain-ayi https://plain-ayi.com/mcp

Cursor: add this to ~/.cursor/mcp.json (or .cursor/mcp.json in a project):

{
  "mcpServers": {
    "plain-ayi": {
      "url": "https://plain-ayi.com/mcp"
    }
  }
}

Prefer plain files? There's a summary for AI tools at /llms.txt and every project's data at /api/projects.json.

Corrections

If you maintain a project listed here, or spot something inaccurate, email ayodejiogunniyi@gmail.com or use the "Report a problem" link on any project page. A person reads every report and fixes mistakes promptly.