Top-AI-Tools: a community link list with no code in it

A directory that contains nothing but links

Top-AI-Tools is a community curated directory whose entire payload is a Markdown document. The repository top level holds four files: README.md, a Chinese version at README.zh-CN.md, an AUTHORS.rst file and the LICENSE. There is no source tree, no package manifest, no test suite and no GitHub release, and the primary language of the repository is listed as unknown, which is consistent with a repository whose content is prose and tables rather than code.

The README describes itself as a community curated directory that aims to catalog the most outstanding AI powered products, and it says plainly that the index is not exhaustive but rather a compilation of the maintainer's research plus contributions from the community. That framing is more useful than it sounds, because it sets the expectation correctly: nothing here has been benchmarked, audited or reviewed on your behalf.

What the visible portion does contain is a themed table of products in an Art and Creativity category. The entries span text to video generators such as Wan3Video, AI music generators such as DoreArt AI, image generators including Midjourney, Craiyon, Playground and DreamStudio, editing tools like igly.ai and PhotoRestore.ai, avatar and headshot generators including AvatarAI, Lensa and Black Headshots, voice work such as VocalReplica, voicetoinstrument.com, Jamorphosia and IRIS Clarity Studio, and transcription or summarisation tools such as Saveto AI, Transcribe Video AI and AI Video Summarizer. Pricing is sometimes captured in a column of its own, with entries marked Free, Free + Premium or a link to a paid page.

The last push was on 2026-10-19, so the list is being edited regularly, which is also the problem: it is edited by many hands with no gate other than a pull request.

The licence is MIT, and the AUTHORS.rst file records contributors.

The three column row format every entry has to follow

Contributions are not free form prose. The README defines a fixed row format, and the header line given for contributors is:

Curated list of top AI Tools.
|Tools|Used For|[🔗](Link)|

In practice the rendered table has three columns, the tool name, a short description of what it is used for, and a link column. The link column is written as a link with a chain glyph in front of the address, which is how a reader is meant to get from the name to the vendor page.

The description column is short by design. An entry is one sentence that says what the product does and, in some cases, whether it is free. Wan3Video is described as an independent browser based Wan 3.0 text to video and image to video generator with prompt guides and model comparisons. igly.ai is a free AI image editor with more than twelve tools for e commerce, covering background removal, generation, inpainting and upscaling. The Lunalisa entry describes an AI creative workspace for product photos, marketing posters and white background listing images, with image to video across thirteen image and twenty two video models.

That constraint is the reason the list is fast to skim and quick to contribute to. It also means the descriptions are vendor claims rather than evaluations, and the list has no field for a licence, an API, a self hosted option, a data retention policy or a review. If you need any of those, the table cannot give them to you.

Duplicate rows, a pricing word in the link column and a truncated entry

The visible portion of the list contains defects that are worth seeing before you rely on it, because they show what happens when a directory is filled in by strangers.

Three entries appear twice. NanoBanana2 is listed once with the name as a link and a separate row with the same description and the link in the link column. ThumblifyAI is repeated in the same pattern, and igly.ai appears twice as well, once as a linked name and once with the link cell reading a closing parenthesis as part of the URL.

The first row of the table is a different kind of broken. The Wan3Video entry has the description in the second column and the third column reads Free + Premium, a pricing phrase where a link belongs. Elsewhere in the list, the pricing information is folded into the third column, as in the DreamStudio row, which puts the Stability AI address in the link position and the open sourced image generation model description in the middle. The same pattern appears with Playground and with the other entries that link to a company site while describing a product in the middle column, which means the table has at least two incompatible conventions for where a price belongs.

And the visible portion ends mid row, on a line that begins with a pipe and the letters Pi, with nothing after it. The document is truncated, so a reader arriving at the tail of the list hits an entry that was never finished.

None of this is fatal for a link list, and all of it is a reason to check the actual vendor page rather than trusting the row. The duplication in particular is the kind of thing that a simple duplicate check on the name column would catch, and the project does not describe one.

Contributions arrive as pull requests, with one promotional row in the way

The contribution path is short. The README asks that you open a pull request for a list of top AI tools to be added, and it tells you to use the format shown. Issues and discussions are not part of the documented flow; the pull request is the mechanism.

There is one detail in that section worth pointing out. Above the contribution instructions sits a single named entry for WebCoreLab AI Audit, described as a 272 check AI powered SEO audit plus GEO and AEO for AI search visibility, and it says it tracks brand citations in ChatGPT, Claude, Perplexity and Gemini. It is the only entry in the contributing section rather than in a themed table, and it appears there with a link to the vendor's own site rather than to a page about the directory.

Whether that is a sponsorship, a favour or simply the first entry someone added is not explained anywhere in the README. What is documented is the placement, and placement matters in a directory whose value is entirely a function of what a reader sees first in a given section.

The rest of the guidance is generic, and the README does not describe an editorial process beyond the pull request. There is no stated policy on how a tool is evaluated before inclusion, no rule about how a vendor pays for placement, no note about whether entries are ever removed when a product is discontinued, and no mention of a review date.

What the directory cannot tell you

The limits are structural rather than accidental, and they are worth being blunt about, because a link list and a curated catalogue are often treated as the same thing.

There is no verification of any kind. The README does not describe a process for checking that a listed product works, that its description is accurate, that the vendor still exists, or that a free tier is still free. Several rows mark pricing, so a reader can see that a product claims a free tier, but nothing in the repository checks that claim or records when it was last true.

There is also no code. If your requirement is an API, a self hosted deployment, a data processing agreement or a licence you can read in the same place as the description, this repository has none of it. The AUTHORS.rst file credits people, and that is the extent of the provenance information available.

Coverage is uneven for the same reason. The visible portion is heavily weighted towards image and video products, and towards consumer facing tools with a free tier. Whether there is a comparable treatment of developer tooling, of enterprise products or of open source models is not visible in what the README shows, and the file is truncated, so the shape of the whole list cannot be judged from the part that is readable.

The Chinese README is a second copy of the same content rather than a translation pipeline. It is a separate file in the repository root, so keeping the two in sync is a manual task, and the README does not say which one is authoritative or who maintains it.

Editorial conclusion

Top-AI-Tools fits a reader who wants a fast, skimmable index of AI image, video, music and transcription products and is willing to check each vendor link themselves. It does not fit anyone who needs a tool with an API, a licence review, a pricing comparison or a maintained catalogue. Verify first that the entry you are about to use has a working link column, since several rows in the visible portion carry a pricing word where the link belongs, and treat the directory as a starting point rather than a shortlist.

Frequently asked questions

Official sources

1. ghimiresunil/Top-AI-Tools on GitHub
2. Issues
3. License: MIT
4. README