xx025/carrot: What the Free ChatGPT Site List Actually Ships

A curated list, not a package you install

The repository has two top-level entries: a .github/ directory and README.md. There is no source tree, no build file, no package manifest. The README describes itself as a list of available AI assistants, agents, developer tools, creative tools and industry applications, and points to a published site at https://cc.ai55.cc/. That is the whole product: a maintained index. The problem it addresses is discovery. Someone looking for a free AI assistant has to sift through search results, dead links and signup walls. This project compresses that search into one page where each entry carries a status marker. The audience is end users choosing a tool, not developers integrating one. If you arrived expecting a library, an SDK, or a self-hosted chat interface, the repository layout says otherwise.

How the README encodes free, login and region status

The README opens with a marker legend rather than prose. Five symbols classify every entry: a new-update flag, a recommendation star, a free-to-use marker, a login-or-password-required marker, and a marker for tools that need international network access. Each entry then appears in an HTML table row with a rank number, a favicon image, a linked name, a short description, and a link cell. The popular section lists twelve entries in that format, starting with ChatGPT, Gemini, Claude, DeepSeek, Doubao, Qianwen, Perplexity, Zhipu Qingyan, Manus, Cursor, Midjourney and Hailuo Video. Further sections group the same style of table by category, for example a general assistants and intelligent search section that repeats ChatGPT, Gemini and Claude. The data flow is therefore manual: a contributor edits README.md, and the published site at cc.ai55.cc renders the same catalog. There is no scraper, no API, and no automated availability check visible in the repository layout. That matters, because the accuracy of every marker depends on someone noticing that a site changed its login policy or went offline.

Reading xx025/carrot for the first time

There is nothing to install. The README gives no install steps, no package name and no runtime. The project says to use the published site at https://cc.ai55.cc/ and to submit sites, report dead entries or correct information through the GitHub Issues page at https://github.com/xx025/carrot/issues. The README itself is the artifact, so the only practical first step is opening it and reading the marker legend before the tables. Start with the legend, then scan the category sections rather than the popular section, because the category tables group comparable tools together. Within a category, read the description column and the marker column in the same pass, so you do not pick a tool that turns out to require a login or international access. When an entry looks wrong, the README points to Issues rather than to a pull request template, so the contribution path runs through issue reports. Nothing in the repository layout suggests a local build, a server, or a configuration file you need to create.

The maintenance model is manual and the markers can go stale

The last push to the default branch was on 2026-09-05, so the repository is not dormant. But a recent push does not mean every row was rechecked. Nothing in the repository layout indicates an automated link checker or a scheduled validation job. The markers for free use, login requirement and international access are human judgments recorded in a Markdown table. A service can move from free to paid, or from open to login-gated, without the table changing. The README itself anticipates this by asking readers to report failed or incorrect entries through Issues, which is a reactive process. Treat every marker as a claim with a timestamp you cannot see per row. The category tables also repeat entries, so ChatGPT, Gemini and Claude appear in both the popular section and the general assistants section. If you edit the file, you have to change each occurrence, and a missed duplicate leaves the page internally inconsistent.

Where a static directory loses to a maintained index or a search engine

The obvious alternative is a general search engine, and the difference is in what you get back. A search for a free AI assistant returns ranked pages with no consistent status vocabulary; you infer free, login and region constraints from each site. This project imposes that vocabulary up front. The trade-off is coverage and freshness: a search engine indexes the entire web continuously, while this list only contains what a contributor added and only stays correct while someone maintains it. A second alternative is an API-based model aggregator, which routes requests to multiple providers behind one interface. That solves a different problem. An aggregator gives you programmatic access and a single credential; this project gives you a page of links and no runtime. If your goal is to call a model from code, an aggregator is the right shape and this repository is not, because it ships no client and no endpoint.

License status and what the repository does not state

No license is stated in the repository, and the README does not include a license section or a link to one. The top-level entries are .github/ and README.md, so there is no LICENSE file at the root. That absence matters more for the text than for the links. The catalog content, meaning the descriptions and the selection itself, has no declared terms for reuse or redistribution. The entries link out to third-party sites, and each of those sites has its own terms, which this project does not summarize. If you plan to mirror the list, republish the tables, or build on the selection, the missing license is the first thing to resolve. This is a description of what the repository states, not legal advice.

What the xx025/carrot list is good for, and what to verify

Use it as a starting shortlist. The marker legend lets you filter quickly: skip entries with the login marker if you want zero friction, and skip entries with the international-access marker if your network cannot reach them. The category sections are more useful than the popular section for that filtering, because the popular section mixes assistants, code editors, image generators and video tools in one table. Do not use it as a dependency. There is no versioned artifact, no release retrieved for this repository, and no API, so nothing here belongs in a build. Before committing to any single tool from the list, open the tool itself and confirm the three things the markers claim: that it is free at the level you need, that it does not require an account you are unwilling to create, and that it loads from your location. The list narrows the search. The verification is still yours.

Editorial conclusion

Adopt xx025/carrot if you want a single browsable page of AI assistants and tools, with free, login-required and international-access flags already marked, and you are willing to verify each entry yourself. Do not adopt it if you need an installable component, an API, or a dependency you can pin in a build. Before relying on it, open the README, check which of the five markers apply to the tools you care about, and confirm the repository still carries a license file, since none is stated in the repository.

Frequently asked questions

Official sources

1. Issues
2. README
3. xx025/carrot on GitHub