# 4664 proxies found, 267 published, and a README that disagrees with itself

> Proxy-List-World is a data repository rather than a tool: four files at the root, no code, and a README that reports what its last scrape found. The numbers are the interesting part, because 4664 proxies discovered becomes 267 published, five of twelve upstream sources return nothing, and the sample table is dominated by a single hosting provider with three duplicated rows. The stated refresh interval also differs between the README and the repository description.

**themiralay/Proxy-List-World** — ICE Proxy list produces new proxies by collectively scanning the entire open source list in the world. Proxies are updated every 2 - 3 minutes. New Proxy sources will be added day by day. The list is always open source.

- Repository: https://github.com/themiralay/Proxy-List-World
- Stars: 185 · Forks: 15
- Language: Unknown
- License: not declared
- Published: 2026-08-21 · Updated: 2026-08-21 · Language: en
- Canonical page: https://hysenlabs.com/projects/themiralay-proxy-list-world

## Four files at the root, and no scraper in the repository

The top level of the repository holds README.md, data.txt, data.json, and data-with-geolocation.json. There is no source directory, no script, no dependency manifest, and no GitHub Actions workflow, and the project has no tagged releases at all. That matters for how you read the rest of the README, because it describes a running process rather than a piece of software you can build. Something scrapes free-proxy sites on a schedule, validates whether each entry works, and rewrites those three files. Whatever performs that work is not versioned here, so the data is reproducible only in the sense that the files are public and the raw URLs are unversioned raw.githubusercontent.com links on the master branch. The last push to the repository is dated September 22, 2026, and it is not archived.

## The README says ten minutes, the description says two to three

Two parts of the repository disagree about how often the list refreshes, and a third label disagrees about what kind of list it is. The README describes a lightweight project that, every 10 minutes, scrapes lots of free-proxy sites, validates if it works, and serves a clean proxy list. The repository description says proxies are updated every 2 to 3 minutes. The same description also opens by calling the project an ICE Proxy list, while the README title is Free HTTP Proxy List and every published file is an HTTP style address and port pair. None of these are fatal, but if you are trying to reason about staleness, the two refresh figures differ by a factor of three and to five, and the protocol label is not consistent between the two places you would look first. The README does not reconcile them or explain which one is current.

## 4664 discovered becomes 267 published, in three formats

The headline line is a scraper status message reading that the scraper found 4664 proxies at the latest update, followed by the note that usable proxies are below. The Usage section then says to click the file format you want and copy the URL, and the table lists three of them, all with a count of 267. data.txt holds ip_address and port combined, one per newline, which is the form most command line tools want. data.json holds ip and port as separate fields. data-with-geolocation.json adds geolocation as a third field. So the published set is a small fraction of what was discovered, and the identical count across all three files is the useful signal here: the geolocation variant is not a partial file, it is the same 267 records with one more column. The README also misspells separated as seperated in the description of the text format, which is a small sign of how little editing this file receives.

## Five of twelve sources return nothing, and one returns most of the list

The Sources table gives two numbers per upstream, how many proxies were found there, and whether the source succeeded. Seven are marked as working: free-proxy-list.net with 300, us-proxy.org with 200, proxylist.geonode.com with 300, proxyscrape.com with 540, clarketm/proxy-list with 400, monosans/proxy-list with 236, and TheSpeedX/PROXY-List with 2688. Five are marked as returning zero: proxydb.net, free-proxy-list.com, proxy-list.download, vpnoverview.com, and proxyscan.io. Two things follow from those numbers. The seven working counts add up to exactly the 4664 in the status line, so the figure is a sum of upstream yields rather than an independent estimate. And a single repository, TheSpeedX, accounts for 2688 of that total, more than half, which means the effective diversity of the scraped pool is narrower than twelve sources suggests.

## The sample table is one hosting provider, and it repeats itself

The twenty sample rows carry ip, port, country, city, and internet service provider, and they are worth reading as a description of the pool rather than as twenty unrelated endpoints. Nine of the twenty are DigitalOcean, spread across the United States, Canada, India, and Singapore, and three of those nine share the 68.183 address block in North Bergen, New Jersey. The rest are spread thinner: Akamai Technologies in Tokyo, LG DACOM in Yongin-si, Microsoft Corporation in Singapore, Amazon Technologies in Ashburn, UCLOUD in Hong Kong, Nocix in Kansas City, Amanah Tech in Toronto, Real Time Data Services in Mumbai, and NetLab Global in Tokyo. Two other patterns show up. Three rows are exact duplicates of earlier rows, 68.183.22.37 on port 10000, 107.150.41.226 on port 18080, and 3.211.120.181 on port 443, so the sample is not deduplicated the way the published files are.

## The upstream queries ask for protocol and anonymity, the output drops them

The geonode source URL in the table requests protocols of http and https and sorts by last checked, and the proxyscrape URL requests a protocol, a timeout, a country of all, an ssl setting, and an anonymity of all. Those are the filters the scraper is asking upstream for. None of them survives into the published output. The three file formats are ip and port, ip and port plus geolocation, and ip and port as text, with no protocol column, no anonymity classification, no latency, and no last-checked timestamp, even though sorting by last checked is exactly the field a consumer would want. The consequence is that a user of data.json cannot tell an HTTP proxy from an HTTPS one, and cannot tell an anonymous proxy from a transparent one, without testing each entry. The geolocation file adds city and provider but no freshness.

## Usage is copy a raw URL, and contributing is one paragraph

The entire usage story is a table of three raw file links on the master branch, and the entire contribution story is a paragraph saying contributions are welcome, greatly appreciated, that every little bit helps, and that credit will always be given. There is no issue template, no pull request guidance, no CODE_OF_CONDUCT, and no CONTRIBUTING document. The description claims the list is always open source, and there is no LICENSE file in the tree, so that claim has no text attached to it and the repository metadata reports no recognised license either. What the README does not state is just as notable: there is no retention policy for an individual entry, no per-record health check result, no fallback behaviour if a source is down, and no statement of what happens to a proxy that worked at the last scrape and has since died. Treat the file as a snapshot with a timestamp you have to infer from the commit history.

## Conclusion

Proxy-List-World is useful as a quick reference for a one-off request and as a worked example of what a scrape-and-validate pipeline publishes, not as a dependency for anything that needs uptime. Before wiring it into a tool, accept that the published files carry no protocol or anonymity field, that much of the sample sits on one hosting provider, and that the refresh interval is stated in two different places. Note also that there is no license file, only four files at the root.

## FAQ

### Where can I find proxy lists?

This repository publishes three formats on its master branch, and the README also names twelve upstream sources it scrapes, including free-proxy-list.net, us-proxy.org, proxyscrape.com, proxylist.geonode.com, and three GitHub proxy-list repositories. Seven of the twelve are marked as succeeding.

### How to use a proxy list?

The README's instruction is to click the file format you want and copy the URL. data.txt gives an ip address and port combined, one per newline, data.json gives ip and port as separate fields, and data-with-geolocation.json adds geolocation. All three files are listed with the same count of 267.

### What does the Proxy-List-World scraper actually do?

It scrapes free-proxy sites on a schedule, validates whether each proxy works, and publishes only the usable ones as three files. The README says the cycle runs every 10 minutes, while the repository description says proxies are updated every 2 to 3 minutes.

### How many proxies does Proxy-List-World publish?

The status line says the scraper found 4664 proxies at the latest update, and each of the three published files lists 267. The per-source counts in the table add up to 4664, with five of the twelve sources marked as returning nothing.

### Is there a license file on Proxy-List-World?

No. The repository root holds only README.md, data.txt, data.json, and data-with-geolocation.json, and the repository metadata reports no recognised license. The description says the list is always open source, but no license text appears in the tree.

## Sources

- [Official README](https://github.com/themiralay/Proxy-List-World#readme)
- [Project repository](https://github.com/themiralay/Proxy-List-World)
- [README](https://github.com/themiralay/Proxy-List-World/blob/master/README.md)
- [Releases](https://github.com/themiralay/Proxy-List-World/releases)

---

Hysen Labs editorial analysis, written from the project's own repository and release notes. Cite the canonical page: https://hysenlabs.com/projects/themiralay-proxy-list-world