proxy-scraper-cli 1.22.0__tar.gz → 1.23.0__tar.gz

{proxy_scraper_cli-1.22.0 → proxy_scraper_cli-1.23.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: proxy-scraper-cli
-Version: 1.22.0
+Version: 1.23.0
 Summary: Finds free HTTP, SOCKS4 and SOCKS5 proxies that actually work: 700+ sources, real checks, honeypot filtering and a rotating local proxy server.
 Author: Maximilian Feix
 License-Expression: MIT
@@ -22,6 +22,7 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
 Classifier: Topic :: Internet :: Proxy Servers
 Classifier: Topic :: System :: Networking
 Requires-Python: >=3.9
@@ -36,7 +37,7 @@ Requires-Dist: mcp<3,>=2.2; python_version >= "3.10" and extra == "mcp"
 Provides-Extra: dev
 Requires-Dist: pytest; extra == "dev"
 Requires-Dist: hypothesis; extra == "dev"
-Requires-Dist: ruff==0.16.8; extra == "dev"
+Requires-Dist: ruff==0.16.9; extra == "dev"
 Dynamic: license-file
 
 <div align="center">
@@ -51,16 +52,19 @@ Dynamic: license-file
 [![Release](https://img.shields.io/github/v/release/maximilianfeix/proxy-scraper?style=flat-square&color=D4F77A&labelColor=121113)](https://github.com/maximilianfeix/proxy-scraper/releases/latest)
 [![Live proxies](https://img.shields.io/endpoint?url=https%3A%2F%2Fraw.githubusercontent.com%2Fmaximilianfeix%2Fproxy-scraper%2Fproxy-list%2Fbadges%2Ftotal.json&style=flat-square&labelColor=121113)](#live-list)
 [![PyPI](https://img.shields.io/pypi/v/proxy-scraper-cli?style=flat-square&color=D4F77A&labelColor=121113&label=pypi)](https://pypi.org/project/proxy-scraper-cli/)
-[![Python](https://img.shields.io/badge/python-3.9–3.13-D4F77A?style=flat-square&labelColor=121113)](pyproject.toml)
+[![Python](https://img.shields.io/badge/python-3.9–3.14-D4F77A?style=flat-square&labelColor=121113)](pyproject.toml)
 [![Stars](https://img.shields.io/github/stars/maximilianfeix/proxy-scraper?style=flat-square&color=D4F77A&labelColor=121113)](https://github.com/maximilianfeix/proxy-scraper/stargazers)
 [![License](https://img.shields.io/badge/license-MIT-D4F77A?style=flat-square&labelColor=121113)](LICENSE)
 
 <a href="https://maximilianfeix.github.io/proxy-scraper/"><img src="https://img.shields.io/badge/Browse_the_live_list-D4F77A?style=for-the-badge&labelColor=121113" alt="Browse the live list"></a>
+<a href="https://github.com/maximilianfeix/free-proxy-list"><img src="https://img.shields.io/badge/Just_the_lists-121113?style=for-the-badge" alt="Just the lists: free-proxy-list"></a>
 <a href="#install"><img src="https://img.shields.io/badge/Install-121113?style=for-the-badge" alt="Install"></a>
 <a href="#from-python"><img src="https://img.shields.io/badge/Python_API-121113?style=for-the-badge" alt="Python API"></a>
 <a href="#mcp"><img src="https://img.shields.io/badge/MCP_server-121113?style=for-the-badge" alt="MCP server for AI agents"></a>
 <a href="bot/"><img src="https://img.shields.io/badge/Discord_bot-121113?style=for-the-badge" alt="Discord bot"></a>
 
+**English** · [简体中文](README.zh-CN.md)
+
 [Install](#install) · [Live list](#live-list) · [For AI agents](#mcp) · [Proxy server](#proxy-server) · [How it works](#how-it-works) · [Recipes](#recipes) · [Options](#options) · [FAQ](#faq)
 
 </div>
@@ -114,6 +118,14 @@ proxy-scraper
 
 The package is called `proxy-scraper-cli` on PyPI (the plain name is taken), the command is `proxy-scraper`.
 
+**With [Homebrew](https://brew.sh)** on macOS and Linux:
+
+```bash
+brew install maximilianfeix/tap/proxy-scraper
+```
+
+The formula follows each release within about a day.
+
 <details>
 <summary><b>Other ways: Docker, pip, a faster event loop, or straight from the repo</b></summary>
 <br>
@@ -178,7 +190,9 @@ Keys: <kbd>↑</kbd><kbd>↓</kbd> select · <kbd>Space</kbd> toggle · <kbd>1</
 
 ## Live proxy list
 
-Don't want to scan yourself? Every hour **GitHub Actions** runs the tool and publishes the hits to the [`proxy-list`](../../tree/proxy-list) branch – every entry worked in the last run, fastest first.
+Don't want to scan yourself? Every hour **GitHub Actions** runs the tool and publishes the hits to the [`proxy-list`](../../tree/proxy-list) branch – every entry worked in the last run, best first: fast ones that are likely to still be up come on top.
+
+> **Just want the lists?** They also live in a repository of their own, **[maximilianfeix/free-proxy-list](https://github.com/maximilianfeix/free-proxy-list)** – one list per protocol, per country and per site, updated every hour, with the numbers of the last run on the front page. Watch or star it to keep it at hand.
 
 <a href="https://maximilianfeix.github.io/proxy-scraper/"><picture>
   <source media="(prefers-color-scheme: dark)" srcset="https://maximilianfeix.github.io/proxy-scraper/chart-dark.svg">
@@ -319,7 +333,7 @@ Started without arguments, the tool asks what you need using the arrow keys –
 <td valign="top">
 
 **Real verification**<br>
-Every hit has to fetch two independent pages – that weeds out **honeypots** that only answer check requests (in some runs 5 out of 6 “hits”). A third request catches proxies that **tamper with content**: in our measurements one in five working proxies injected a script into a plain HTML page. Plus: HTTPS through a tunnel with **verified TLS**, anonymity level *elite / anonymous / transparent* and the country of the exit IP.
+Every hit has to fetch two independent pages – that weeds out **honeypots** that only answer check requests: over 3.6 million checks, 54 % of the proxies that answered the first request failed the second. A third request catches proxies that **tamper with content**: about one in 40 working proxies changed a known page, usually by injecting a script ([the numbers](docs/free-proxies-in-numbers.md)). Plus: HTTPS through a tunnel with **verified TLS**, anonymity level *elite / anonymous / transparent* and the country of the exit IP.
 
 </td>
 <td valign="top">
@@ -361,13 +375,13 @@ Speed chart, latency histogram, protocols, countries and the latest hits in real
 <td valign="top">
 
 **Runs everywhere**<br>
-macOS, Linux and Windows, Python 3.9 to 3.13. Only two dependencies: `rich` and `certifi`. It even notices when a firewall blocks proxies.
+macOS, Linux and Windows, Python 3.9 to 3.14. Only two dependencies: `rich` and `certifi`. It even notices when a firewall blocks proxies.
 
 </td>
 <td valign="top">
 
 **Thoroughly tested**<br>
-450+ tests run offline against real mini proxies and honeypots on `localhost` – on Linux, macOS and Windows with Python 3.9, 3.11 and 3.13.
+760+ tests run offline against real mini proxies and honeypots on `localhost` – on Linux, macOS and Windows with Python 3.9, 3.11, 3.13 and 3.14.
 
 </td>
 </tr>
@@ -515,7 +529,7 @@ Without a password, `--serve-host` means **anyone who reaches the port can use i
 
 - every connection goes through a different proxy (unless sticky); fast and proven ones are preferred
 - `CONNECT` for HTTPS and plain HTTP requests; HTTP, SOCKS4 and SOCKS5 proxies can sit behind it (SOCKS5 with DNS through the proxy)
-- HTTPS only uses proxies that passed the test with **verified TLS** – no broken encryption
+- HTTPS prefers proxies that passed the test with **verified TLS**; when no such proxy fits right now – none matches a wish like `country-de`, or they have all dropped out – it uses the others – your client's own certificate check still catches a proxy that breaks the encryption, so leave it on
 - if a proxy stays silent inside the tunnel or returns an error page instead of TLS, the same first packet quietly goes to the next one
 - three failures in a row and a proxy leaves the rotation – every 5 minutes those get re-checked and come back if they work again
 - `--serve-refill 6` checks fresh proxies every 6 hours in the background (the live list with `--recheck live`, otherwise the last run + history) with the same checks and filters, and adds the hits – a server that runs for days doesn't run dry
@@ -523,6 +537,10 @@ Without a password, `--serve-host` means **anyone who reaches the port can use i
 
 In testing: 20 of 20 HTTPS requests succeeded, over 15 different exit IPs. In the wizard this is **Proxy server right away**.
 
+**Live dashboard:** open `http://127.0.0.1:8899/__proxy-scraper/` in a browser – the pool, success rate, traffic, a live chart of requests, countries, the best proxies and the latest connections, refreshed every two seconds. It needs nothing from the internet, and `--serve-password` protects it like the rest.
+
+<div align="center"><img src="docs/dashboard-live.png" alt="The live dashboard of the rotating proxy server: usable proxies, success rate, requests per second, countries and the latest connections" width="860"></div>
+
 <a id="pool-api"></a>
 
 ### Proxy pool API
@@ -741,6 +759,18 @@ proxychains4 -f results/latest/proxychains.conf curl https://api.ipify.org
 
 `clash.yaml` has all HTTP and SOCKS5 proxies plus a `url-test` group that always picks the fastest. `singbox.json` does the same for sing-box, SOCKS4 included, and opens a local proxy: `sing-box run -c results/latest/singbox.json`, then use `127.0.0.1:2080` as HTTP or SOCKS5 proxy. All three leave out HTTP proxies that can't tunnel (`CONNECT`), because these tools tunnel everything.
 
+**Browsers and the OS** – `--export pac` writes `proxy.pac`: point Firefox, Chrome (via the system settings), FoxyProxy or macOS/Windows proxy settings at it and every request goes through the best 30 HTTP and SOCKS5 proxies in order, the browser moving on by itself when one fails. Local names and private IP addresses stay direct (a host name like `10.example.com` doesn't count as private). The file itself has no `DIRECT` fallback and resolves no host names, and SOCKS4 is left out because it would make the browser resolve names with your own DNS. Two Firefox settings matter: turn on *Proxy DNS when using SOCKS v5*, and set `network.proxy.failover_direct` to `false` in `about:config` – otherwise Firefox goes direct once every proxy has failed. No install at all: the live list publishes one every hour at `https://maximilianfeix.github.io/proxy-scraper/proxy.pac`.
+
+**Burp Suite, OWASP ZAP, mitmproxy** – put the rotating server behind your intercepting proxy, so a scan or a brute-force test comes from many IPs while you still see every request. Start it with `proxy-scraper --recheck live --serve`, then:
+
+| Tool | Where | Set |
+|---|---|---|
+| Burp Suite | *Settings → Network → Connections → Upstream proxy servers* | a rule for destination host `*`, proxy host `127.0.0.1`, port `8899`; with `--serve-password` add Basic auth (user `any` or a wish like `country-de`, the password) |
+| OWASP ZAP | *Options → Network → Connection → HTTP Proxy* | host `127.0.0.1`, port `8899`, and under its authentication the user (`any` or a wish) and the password – use the HTTP setting, not SOCKS: ZAP never sends loopback addresses through its SOCKS proxy |
+| mitmproxy | command line | `mitmproxy --mode upstream:http://127.0.0.1:8899 --upstream-auth country-us:x` |
+
+Username wishes – `country-us` sends everything out through US proxies, `session-NAME` keeps one exit IP for a login flow – reach the server as the proxy login. mitmproxy sends it right away; Burp and ZAP may wait to be asked, so start the server with `--serve-password` when you rely on a wish there. **Keep your tool's upstream certificate checks on** (no `--ssl-insecure` in mitmproxy, no "ignore certificate errors" upstream in Burp or ZAP): free proxies are strangers, and that check is what stops one from reading or changing your HTTPS traffic. Tested end to end with mitmproxy 12, certificate checks on: HTTP and HTTPS through mitmproxy and the rotating server, every request out through a US proxy, and a site with an expired certificate refused. Only test what you're allowed to.
+
 **In a pipe** – `-o -` prints the hits to stdout, the interface moves to stderr
 
 ```bash
@@ -834,7 +864,7 @@ curl.exe -x (Get-Content "$run\all.txt" -TotalCount 1) http://api.ipify.org
 | `--serve-refill HOURS` | while serving, check fresh proxies every HOURS and add the hits to the pool |
 | `--serve [PORT]` | afterwards serve as a rotating proxy on `127.0.0.1:PORT` (default: 8899) |
 | `-o FILE` | also write all hits to this file; `-o -` prints them to stdout |
-| `--export FORMATS` | extra files for other tools: `proxychains`, `clash`, `singbox`, `curl` or `all` |
+| `--export FORMATS` | extra files for other tools: `proxychains`, `clash`, `singbox`, `curl`, `pac` or `all` |
 | `-V`, `--version` | print the version |
 | `--completion SHELL` | print the tab completion script for bash, zsh, fish or PowerShell |
 
@@ -920,7 +950,7 @@ Only for things that don't matter. Public proxies are run by strangers who can r
 
 What's next is in the [open issues](../../issues) – ideas and wishes are welcome as an [issue](../../issues/new/choose). During Hacktoberfest there are [beginner-friendly issues](../../issues?q=is%3Aopen+label%3Ahacktoberfest) with pointers on where to start.
 
-- [ ] [Mirror the daily snapshots to a Hugging Face dataset](../../issues/188)
+Shipped in v1.22 and v1.23: a live dashboard for the rotating server, `--export pac` and an hourly `proxy.pac`, "best first" that counts how likely a proxy is still up, the lists in a repository of their own ([free-proxy-list](https://github.com/maximilianfeix/free-proxy-list)), daily snapshots on Hugging Face, `brew install`, a Chinese README, Python 3.14, and a recipe for Burp Suite, ZAP and mitmproxy.
 
 Shipped in v1.10 to v1.21: download speed per proxy and "best first" everywhere, daily snapshots, a rotating proxy URL for requests, httpx and Playwright, `--pick` for working proxies without a scan, a proxy pool API compatible with jhao104/proxy_pool, your own lists with `--source`, a GitHub Action, a Telegram bot and an Atom feed for the weekly report.
 
@@ -969,12 +999,13 @@ proxyscraper/
 ├── geodb.py            DB-IP country database (monthly, binary search)
 ├── targets.py          target sites for --target
 ├── output.py           result files
-├── exporters.py        proxychains, Clash, sing-box and curl formats (--export)
+├── exporters.py        proxychains, Clash, sing-box, curl and PAC formats (--export)
 ├── server/             rotating proxy server (--serve): pool · http · upstream · socks · status · api · core
 ├── api.py              find_proxies() / check_proxies() for Python
 ├── agent.py            MCP tools without the SDK: live list, filters, fetch through proxies
 ├── mcp_server.py       MCP server (proxy-scraper-mcp) for AI agents
 ├── publish.py          live list for GitHub Actions
+├── mirror.py           the same lists in their own repo (free-proxy-list)
 ├── paths.py            where state and results are stored
 ├── compat.py           differences between Unix and Windows
 ├── netio.py            small HTTP client on asyncio

{proxy_scraper_cli-1.22.0 → proxy_scraper_cli-1.23.0}/README.md

@@ -10,16 +10,19 @@
 [![Release](https://img.shields.io/github/v/release/maximilianfeix/proxy-scraper?style=flat-square&color=D4F77A&labelColor=121113)](https://github.com/maximilianfeix/proxy-scraper/releases/latest)
 [![Live proxies](https://img.shields.io/endpoint?url=https%3A%2F%2Fraw.githubusercontent.com%2Fmaximilianfeix%2Fproxy-scraper%2Fproxy-list%2Fbadges%2Ftotal.json&style=flat-square&labelColor=121113)](#live-list)
 [![PyPI](https://img.shields.io/pypi/v/proxy-scraper-cli?style=flat-square&color=D4F77A&labelColor=121113&label=pypi)](https://pypi.org/project/proxy-scraper-cli/)
-[![Python](https://img.shields.io/badge/python-3.9–3.13-D4F77A?style=flat-square&labelColor=121113)](pyproject.toml)
+[![Python](https://img.shields.io/badge/python-3.9–3.14-D4F77A?style=flat-square&labelColor=121113)](pyproject.toml)
 [![Stars](https://img.shields.io/github/stars/maximilianfeix/proxy-scraper?style=flat-square&color=D4F77A&labelColor=121113)](https://github.com/maximilianfeix/proxy-scraper/stargazers)
 [![License](https://img.shields.io/badge/license-MIT-D4F77A?style=flat-square&labelColor=121113)](LICENSE)
 
 <a href="https://maximilianfeix.github.io/proxy-scraper/"><img src="https://img.shields.io/badge/Browse_the_live_list-D4F77A?style=for-the-badge&labelColor=121113" alt="Browse the live list"></a>
+<a href="https://github.com/maximilianfeix/free-proxy-list"><img src="https://img.shields.io/badge/Just_the_lists-121113?style=for-the-badge" alt="Just the lists: free-proxy-list"></a>
 <a href="#install"><img src="https://img.shields.io/badge/Install-121113?style=for-the-badge" alt="Install"></a>
 <a href="#from-python"><img src="https://img.shields.io/badge/Python_API-121113?style=for-the-badge" alt="Python API"></a>
 <a href="#mcp"><img src="https://img.shields.io/badge/MCP_server-121113?style=for-the-badge" alt="MCP server for AI agents"></a>
 <a href="bot/"><img src="https://img.shields.io/badge/Discord_bot-121113?style=for-the-badge" alt="Discord bot"></a>
 
+**English** · [简体中文](README.zh-CN.md)
+
 [Install](#install) · [Live list](#live-list) · [For AI agents](#mcp) · [Proxy server](#proxy-server) · [How it works](#how-it-works) · [Recipes](#recipes) · [Options](#options) · [FAQ](#faq)
 
 </div>
@@ -73,6 +76,14 @@ proxy-scraper
 
 The package is called `proxy-scraper-cli` on PyPI (the plain name is taken), the command is `proxy-scraper`.
 
+**With [Homebrew](https://brew.sh)** on macOS and Linux:
+
+```bash
+brew install maximilianfeix/tap/proxy-scraper
+```
+
+The formula follows each release within about a day.
+
 <details>
 <summary><b>Other ways: Docker, pip, a faster event loop, or straight from the repo</b></summary>
 <br>
@@ -137,7 +148,9 @@ Keys: <kbd>↑</kbd><kbd>↓</kbd> select · <kbd>Space</kbd> toggle · <kbd>1</
 
 ## Live proxy list
 
-Don't want to scan yourself? Every hour **GitHub Actions** runs the tool and publishes the hits to the [`proxy-list`](../../tree/proxy-list) branch – every entry worked in the last run, fastest first.
+Don't want to scan yourself? Every hour **GitHub Actions** runs the tool and publishes the hits to the [`proxy-list`](../../tree/proxy-list) branch – every entry worked in the last run, best first: fast ones that are likely to still be up come on top.
+
+> **Just want the lists?** They also live in a repository of their own, **[maximilianfeix/free-proxy-list](https://github.com/maximilianfeix/free-proxy-list)** – one list per protocol, per country and per site, updated every hour, with the numbers of the last run on the front page. Watch or star it to keep it at hand.
 
 <a href="https://maximilianfeix.github.io/proxy-scraper/"><picture>
   <source media="(prefers-color-scheme: dark)" srcset="https://maximilianfeix.github.io/proxy-scraper/chart-dark.svg">
@@ -278,7 +291,7 @@ Started without arguments, the tool asks what you need using the arrow keys –
 <td valign="top">
 
 **Real verification**<br>
-Every hit has to fetch two independent pages – that weeds out **honeypots** that only answer check requests (in some runs 5 out of 6 “hits”). A third request catches proxies that **tamper with content**: in our measurements one in five working proxies injected a script into a plain HTML page. Plus: HTTPS through a tunnel with **verified TLS**, anonymity level *elite / anonymous / transparent* and the country of the exit IP.
+Every hit has to fetch two independent pages – that weeds out **honeypots** that only answer check requests: over 3.6 million checks, 54 % of the proxies that answered the first request failed the second. A third request catches proxies that **tamper with content**: about one in 40 working proxies changed a known page, usually by injecting a script ([the numbers](docs/free-proxies-in-numbers.md)). Plus: HTTPS through a tunnel with **verified TLS**, anonymity level *elite / anonymous / transparent* and the country of the exit IP.
 
 </td>
 <td valign="top">
@@ -320,13 +333,13 @@ Speed chart, latency histogram, protocols, countries and the latest hits in real
 <td valign="top">
 
 **Runs everywhere**<br>
-macOS, Linux and Windows, Python 3.9 to 3.13. Only two dependencies: `rich` and `certifi`. It even notices when a firewall blocks proxies.
+macOS, Linux and Windows, Python 3.9 to 3.14. Only two dependencies: `rich` and `certifi`. It even notices when a firewall blocks proxies.
 
 </td>
 <td valign="top">
 
 **Thoroughly tested**<br>
-450+ tests run offline against real mini proxies and honeypots on `localhost` – on Linux, macOS and Windows with Python 3.9, 3.11 and 3.13.
+760+ tests run offline against real mini proxies and honeypots on `localhost` – on Linux, macOS and Windows with Python 3.9, 3.11, 3.13 and 3.14.
 
 </td>
 </tr>
@@ -474,7 +487,7 @@ Without a password, `--serve-host` means **anyone who reaches the port can use i
 
 - every connection goes through a different proxy (unless sticky); fast and proven ones are preferred
 - `CONNECT` for HTTPS and plain HTTP requests; HTTP, SOCKS4 and SOCKS5 proxies can sit behind it (SOCKS5 with DNS through the proxy)
-- HTTPS only uses proxies that passed the test with **verified TLS** – no broken encryption
+- HTTPS prefers proxies that passed the test with **verified TLS**; when no such proxy fits right now – none matches a wish like `country-de`, or they have all dropped out – it uses the others – your client's own certificate check still catches a proxy that breaks the encryption, so leave it on
 - if a proxy stays silent inside the tunnel or returns an error page instead of TLS, the same first packet quietly goes to the next one
 - three failures in a row and a proxy leaves the rotation – every 5 minutes those get re-checked and come back if they work again
 - `--serve-refill 6` checks fresh proxies every 6 hours in the background (the live list with `--recheck live`, otherwise the last run + history) with the same checks and filters, and adds the hits – a server that runs for days doesn't run dry
@@ -482,6 +495,10 @@ Without a password, `--serve-host` means **anyone who reaches the port can use i
 
 In testing: 20 of 20 HTTPS requests succeeded, over 15 different exit IPs. In the wizard this is **Proxy server right away**.
 
+**Live dashboard:** open `http://127.0.0.1:8899/__proxy-scraper/` in a browser – the pool, success rate, traffic, a live chart of requests, countries, the best proxies and the latest connections, refreshed every two seconds. It needs nothing from the internet, and `--serve-password` protects it like the rest.
+
+<div align="center"><img src="docs/dashboard-live.png" alt="The live dashboard of the rotating proxy server: usable proxies, success rate, requests per second, countries and the latest connections" width="860"></div>
+
 <a id="pool-api"></a>
 
 ### Proxy pool API
@@ -700,6 +717,18 @@ proxychains4 -f results/latest/proxychains.conf curl https://api.ipify.org
 
 `clash.yaml` has all HTTP and SOCKS5 proxies plus a `url-test` group that always picks the fastest. `singbox.json` does the same for sing-box, SOCKS4 included, and opens a local proxy: `sing-box run -c results/latest/singbox.json`, then use `127.0.0.1:2080` as HTTP or SOCKS5 proxy. All three leave out HTTP proxies that can't tunnel (`CONNECT`), because these tools tunnel everything.
 
+**Browsers and the OS** – `--export pac` writes `proxy.pac`: point Firefox, Chrome (via the system settings), FoxyProxy or macOS/Windows proxy settings at it and every request goes through the best 30 HTTP and SOCKS5 proxies in order, the browser moving on by itself when one fails. Local names and private IP addresses stay direct (a host name like `10.example.com` doesn't count as private). The file itself has no `DIRECT` fallback and resolves no host names, and SOCKS4 is left out because it would make the browser resolve names with your own DNS. Two Firefox settings matter: turn on *Proxy DNS when using SOCKS v5*, and set `network.proxy.failover_direct` to `false` in `about:config` – otherwise Firefox goes direct once every proxy has failed. No install at all: the live list publishes one every hour at `https://maximilianfeix.github.io/proxy-scraper/proxy.pac`.
+
+**Burp Suite, OWASP ZAP, mitmproxy** – put the rotating server behind your intercepting proxy, so a scan or a brute-force test comes from many IPs while you still see every request. Start it with `proxy-scraper --recheck live --serve`, then:
+
+| Tool | Where | Set |
+|---|---|---|
+| Burp Suite | *Settings → Network → Connections → Upstream proxy servers* | a rule for destination host `*`, proxy host `127.0.0.1`, port `8899`; with `--serve-password` add Basic auth (user `any` or a wish like `country-de`, the password) |
+| OWASP ZAP | *Options → Network → Connection → HTTP Proxy* | host `127.0.0.1`, port `8899`, and under its authentication the user (`any` or a wish) and the password – use the HTTP setting, not SOCKS: ZAP never sends loopback addresses through its SOCKS proxy |
+| mitmproxy | command line | `mitmproxy --mode upstream:http://127.0.0.1:8899 --upstream-auth country-us:x` |
+
+Username wishes – `country-us` sends everything out through US proxies, `session-NAME` keeps one exit IP for a login flow – reach the server as the proxy login. mitmproxy sends it right away; Burp and ZAP may wait to be asked, so start the server with `--serve-password` when you rely on a wish there. **Keep your tool's upstream certificate checks on** (no `--ssl-insecure` in mitmproxy, no "ignore certificate errors" upstream in Burp or ZAP): free proxies are strangers, and that check is what stops one from reading or changing your HTTPS traffic. Tested end to end with mitmproxy 12, certificate checks on: HTTP and HTTPS through mitmproxy and the rotating server, every request out through a US proxy, and a site with an expired certificate refused. Only test what you're allowed to.
+
 **In a pipe** – `-o -` prints the hits to stdout, the interface moves to stderr
 
 ```bash
@@ -793,7 +822,7 @@ curl.exe -x (Get-Content "$run\all.txt" -TotalCount 1) http://api.ipify.org
 | `--serve-refill HOURS` | while serving, check fresh proxies every HOURS and add the hits to the pool |
 | `--serve [PORT]` | afterwards serve as a rotating proxy on `127.0.0.1:PORT` (default: 8899) |
 | `-o FILE` | also write all hits to this file; `-o -` prints them to stdout |
-| `--export FORMATS` | extra files for other tools: `proxychains`, `clash`, `singbox`, `curl` or `all` |
+| `--export FORMATS` | extra files for other tools: `proxychains`, `clash`, `singbox`, `curl`, `pac` or `all` |
 | `-V`, `--version` | print the version |
 | `--completion SHELL` | print the tab completion script for bash, zsh, fish or PowerShell |
 
@@ -879,7 +908,7 @@ Only for things that don't matter. Public proxies are run by strangers who can r
 
 What's next is in the [open issues](../../issues) – ideas and wishes are welcome as an [issue](../../issues/new/choose). During Hacktoberfest there are [beginner-friendly issues](../../issues?q=is%3Aopen+label%3Ahacktoberfest) with pointers on where to start.
 
-- [ ] [Mirror the daily snapshots to a Hugging Face dataset](../../issues/188)
+Shipped in v1.22 and v1.23: a live dashboard for the rotating server, `--export pac` and an hourly `proxy.pac`, "best first" that counts how likely a proxy is still up, the lists in a repository of their own ([free-proxy-list](https://github.com/maximilianfeix/free-proxy-list)), daily snapshots on Hugging Face, `brew install`, a Chinese README, Python 3.14, and a recipe for Burp Suite, ZAP and mitmproxy.
 
 Shipped in v1.10 to v1.21: download speed per proxy and "best first" everywhere, daily snapshots, a rotating proxy URL for requests, httpx and Playwright, `--pick` for working proxies without a scan, a proxy pool API compatible with jhao104/proxy_pool, your own lists with `--source`, a GitHub Action, a Telegram bot and an Atom feed for the weekly report.
 
@@ -928,12 +957,13 @@ proxyscraper/
 ├── geodb.py            DB-IP country database (monthly, binary search)
 ├── targets.py          target sites for --target
 ├── output.py           result files
-├── exporters.py        proxychains, Clash, sing-box and curl formats (--export)
+├── exporters.py        proxychains, Clash, sing-box, curl and PAC formats (--export)
 ├── server/             rotating proxy server (--serve): pool · http · upstream · socks · status · api · core
 ├── api.py              find_proxies() / check_proxies() for Python
 ├── agent.py            MCP tools without the SDK: live list, filters, fetch through proxies
 ├── mcp_server.py       MCP server (proxy-scraper-mcp) for AI agents
 ├── publish.py          live list for GitHub Actions
+├── mirror.py           the same lists in their own repo (free-proxy-list)
 ├── paths.py            where state and results are stored
 ├── compat.py           differences between Unix and Windows
 ├── netio.py            small HTTP client on asyncio

{proxy_scraper_cli-1.22.0 → proxy_scraper_cli-1.23.0}/proxy_scraper_cli.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: proxy-scraper-cli
-Version: 1.22.0
+Version: 1.23.0
 Summary: Finds free HTTP, SOCKS4 and SOCKS5 proxies that actually work: 700+ sources, real checks, honeypot filtering and a rotating local proxy server.
 Author: Maximilian Feix
 License-Expression: MIT
@@ -22,6 +22,7 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Programming Language :: Python :: 3.13
+Classifier: Programming Language :: Python :: 3.14
 Classifier: Topic :: Internet :: Proxy Servers
 Classifier: Topic :: System :: Networking
 Requires-Python: >=3.9
@@ -36,7 +37,7 @@ Requires-Dist: mcp<3,>=2.2; python_version >= "3.10" and extra == "mcp"
 Provides-Extra: dev
 Requires-Dist: pytest; extra == "dev"
 Requires-Dist: hypothesis; extra == "dev"
-Requires-Dist: ruff==0.16.8; extra == "dev"
+Requires-Dist: ruff==0.16.9; extra == "dev"
 Dynamic: license-file
 
 <div align="center">
@@ -51,16 +52,19 @@ Dynamic: license-file
 [![Release](https://img.shields.io/github/v/release/maximilianfeix/proxy-scraper?style=flat-square&color=D4F77A&labelColor=121113)](https://github.com/maximilianfeix/proxy-scraper/releases/latest)
 [![Live proxies](https://img.shields.io/endpoint?url=https%3A%2F%2Fraw.githubusercontent.com%2Fmaximilianfeix%2Fproxy-scraper%2Fproxy-list%2Fbadges%2Ftotal.json&style=flat-square&labelColor=121113)](#live-list)
 [![PyPI](https://img.shields.io/pypi/v/proxy-scraper-cli?style=flat-square&color=D4F77A&labelColor=121113&label=pypi)](https://pypi.org/project/proxy-scraper-cli/)
-[![Python](https://img.shields.io/badge/python-3.9–3.13-D4F77A?style=flat-square&labelColor=121113)](pyproject.toml)
+[![Python](https://img.shields.io/badge/python-3.9–3.14-D4F77A?style=flat-square&labelColor=121113)](pyproject.toml)
 [![Stars](https://img.shields.io/github/stars/maximilianfeix/proxy-scraper?style=flat-square&color=D4F77A&labelColor=121113)](https://github.com/maximilianfeix/proxy-scraper/stargazers)
 [![License](https://img.shields.io/badge/license-MIT-D4F77A?style=flat-square&labelColor=121113)](LICENSE)
 
 <a href="https://maximilianfeix.github.io/proxy-scraper/"><img src="https://img.shields.io/badge/Browse_the_live_list-D4F77A?style=for-the-badge&labelColor=121113" alt="Browse the live list"></a>
+<a href="https://github.com/maximilianfeix/free-proxy-list"><img src="https://img.shields.io/badge/Just_the_lists-121113?style=for-the-badge" alt="Just the lists: free-proxy-list"></a>
 <a href="#install"><img src="https://img.shields.io/badge/Install-121113?style=for-the-badge" alt="Install"></a>
 <a href="#from-python"><img src="https://img.shields.io/badge/Python_API-121113?style=for-the-badge" alt="Python API"></a>
 <a href="#mcp"><img src="https://img.shields.io/badge/MCP_server-121113?style=for-the-badge" alt="MCP server for AI agents"></a>
 <a href="bot/"><img src="https://img.shields.io/badge/Discord_bot-121113?style=for-the-badge" alt="Discord bot"></a>
 
+**English** · [简体中文](README.zh-CN.md)
+
 [Install](#install) · [Live list](#live-list) · [For AI agents](#mcp) · [Proxy server](#proxy-server) · [How it works](#how-it-works) · [Recipes](#recipes) · [Options](#options) · [FAQ](#faq)
 
 </div>
@@ -114,6 +118,14 @@ proxy-scraper
 
 The package is called `proxy-scraper-cli` on PyPI (the plain name is taken), the command is `proxy-scraper`.
 
+**With [Homebrew](https://brew.sh)** on macOS and Linux:
+
+```bash
+brew install maximilianfeix/tap/proxy-scraper
+```
+
+The formula follows each release within about a day.
+
 <details>
 <summary><b>Other ways: Docker, pip, a faster event loop, or straight from the repo</b></summary>
 <br>
@@ -178,7 +190,9 @@ Keys: <kbd>↑</kbd><kbd>↓</kbd> select · <kbd>Space</kbd> toggle · <kbd>1</
 
 ## Live proxy list
 
-Don't want to scan yourself? Every hour **GitHub Actions** runs the tool and publishes the hits to the [`proxy-list`](../../tree/proxy-list) branch – every entry worked in the last run, fastest first.
+Don't want to scan yourself? Every hour **GitHub Actions** runs the tool and publishes the hits to the [`proxy-list`](../../tree/proxy-list) branch – every entry worked in the last run, best first: fast ones that are likely to still be up come on top.
+
+> **Just want the lists?** They also live in a repository of their own, **[maximilianfeix/free-proxy-list](https://github.com/maximilianfeix/free-proxy-list)** – one list per protocol, per country and per site, updated every hour, with the numbers of the last run on the front page. Watch or star it to keep it at hand.
 
 <a href="https://maximilianfeix.github.io/proxy-scraper/"><picture>
   <source media="(prefers-color-scheme: dark)" srcset="https://maximilianfeix.github.io/proxy-scraper/chart-dark.svg">
@@ -319,7 +333,7 @@ Started without arguments, the tool asks what you need using the arrow keys –
 <td valign="top">
 
 **Real verification**<br>
-Every hit has to fetch two independent pages – that weeds out **honeypots** that only answer check requests (in some runs 5 out of 6 “hits”). A third request catches proxies that **tamper with content**: in our measurements one in five working proxies injected a script into a plain HTML page. Plus: HTTPS through a tunnel with **verified TLS**, anonymity level *elite / anonymous / transparent* and the country of the exit IP.
+Every hit has to fetch two independent pages – that weeds out **honeypots** that only answer check requests: over 3.6 million checks, 54 % of the proxies that answered the first request failed the second. A third request catches proxies that **tamper with content**: about one in 40 working proxies changed a known page, usually by injecting a script ([the numbers](docs/free-proxies-in-numbers.md)). Plus: HTTPS through a tunnel with **verified TLS**, anonymity level *elite / anonymous / transparent* and the country of the exit IP.
 
 </td>
 <td valign="top">
@@ -361,13 +375,13 @@ Speed chart, latency histogram, protocols, countries and the latest hits in real
 <td valign="top">
 
 **Runs everywhere**<br>
-macOS, Linux and Windows, Python 3.9 to 3.13. Only two dependencies: `rich` and `certifi`. It even notices when a firewall blocks proxies.
+macOS, Linux and Windows, Python 3.9 to 3.14. Only two dependencies: `rich` and `certifi`. It even notices when a firewall blocks proxies.
 
 </td>
 <td valign="top">
 
 **Thoroughly tested**<br>
-450+ tests run offline against real mini proxies and honeypots on `localhost` – on Linux, macOS and Windows with Python 3.9, 3.11 and 3.13.
+760+ tests run offline against real mini proxies and honeypots on `localhost` – on Linux, macOS and Windows with Python 3.9, 3.11, 3.13 and 3.14.
 
 </td>
 </tr>
@@ -515,7 +529,7 @@ Without a password, `--serve-host` means **anyone who reaches the port can use i
 
 - every connection goes through a different proxy (unless sticky); fast and proven ones are preferred
 - `CONNECT` for HTTPS and plain HTTP requests; HTTP, SOCKS4 and SOCKS5 proxies can sit behind it (SOCKS5 with DNS through the proxy)
-- HTTPS only uses proxies that passed the test with **verified TLS** – no broken encryption
+- HTTPS prefers proxies that passed the test with **verified TLS**; when no such proxy fits right now – none matches a wish like `country-de`, or they have all dropped out – it uses the others – your client's own certificate check still catches a proxy that breaks the encryption, so leave it on
 - if a proxy stays silent inside the tunnel or returns an error page instead of TLS, the same first packet quietly goes to the next one
 - three failures in a row and a proxy leaves the rotation – every 5 minutes those get re-checked and come back if they work again
 - `--serve-refill 6` checks fresh proxies every 6 hours in the background (the live list with `--recheck live`, otherwise the last run + history) with the same checks and filters, and adds the hits – a server that runs for days doesn't run dry
@@ -523,6 +537,10 @@ Without a password, `--serve-host` means **anyone who reaches the port can use i
 
 In testing: 20 of 20 HTTPS requests succeeded, over 15 different exit IPs. In the wizard this is **Proxy server right away**.
 
+**Live dashboard:** open `http://127.0.0.1:8899/__proxy-scraper/` in a browser – the pool, success rate, traffic, a live chart of requests, countries, the best proxies and the latest connections, refreshed every two seconds. It needs nothing from the internet, and `--serve-password` protects it like the rest.
+
+<div align="center"><img src="docs/dashboard-live.png" alt="The live dashboard of the rotating proxy server: usable proxies, success rate, requests per second, countries and the latest connections" width="860"></div>
+
 <a id="pool-api"></a>
 
 ### Proxy pool API
@@ -741,6 +759,18 @@ proxychains4 -f results/latest/proxychains.conf curl https://api.ipify.org
 
 `clash.yaml` has all HTTP and SOCKS5 proxies plus a `url-test` group that always picks the fastest. `singbox.json` does the same for sing-box, SOCKS4 included, and opens a local proxy: `sing-box run -c results/latest/singbox.json`, then use `127.0.0.1:2080` as HTTP or SOCKS5 proxy. All three leave out HTTP proxies that can't tunnel (`CONNECT`), because these tools tunnel everything.
 
+**Browsers and the OS** – `--export pac` writes `proxy.pac`: point Firefox, Chrome (via the system settings), FoxyProxy or macOS/Windows proxy settings at it and every request goes through the best 30 HTTP and SOCKS5 proxies in order, the browser moving on by itself when one fails. Local names and private IP addresses stay direct (a host name like `10.example.com` doesn't count as private). The file itself has no `DIRECT` fallback and resolves no host names, and SOCKS4 is left out because it would make the browser resolve names with your own DNS. Two Firefox settings matter: turn on *Proxy DNS when using SOCKS v5*, and set `network.proxy.failover_direct` to `false` in `about:config` – otherwise Firefox goes direct once every proxy has failed. No install at all: the live list publishes one every hour at `https://maximilianfeix.github.io/proxy-scraper/proxy.pac`.
+
+**Burp Suite, OWASP ZAP, mitmproxy** – put the rotating server behind your intercepting proxy, so a scan or a brute-force test comes from many IPs while you still see every request. Start it with `proxy-scraper --recheck live --serve`, then:
+
+| Tool | Where | Set |
+|---|---|---|
+| Burp Suite | *Settings → Network → Connections → Upstream proxy servers* | a rule for destination host `*`, proxy host `127.0.0.1`, port `8899`; with `--serve-password` add Basic auth (user `any` or a wish like `country-de`, the password) |
+| OWASP ZAP | *Options → Network → Connection → HTTP Proxy* | host `127.0.0.1`, port `8899`, and under its authentication the user (`any` or a wish) and the password – use the HTTP setting, not SOCKS: ZAP never sends loopback addresses through its SOCKS proxy |
+| mitmproxy | command line | `mitmproxy --mode upstream:http://127.0.0.1:8899 --upstream-auth country-us:x` |
+
+Username wishes – `country-us` sends everything out through US proxies, `session-NAME` keeps one exit IP for a login flow – reach the server as the proxy login. mitmproxy sends it right away; Burp and ZAP may wait to be asked, so start the server with `--serve-password` when you rely on a wish there. **Keep your tool's upstream certificate checks on** (no `--ssl-insecure` in mitmproxy, no "ignore certificate errors" upstream in Burp or ZAP): free proxies are strangers, and that check is what stops one from reading or changing your HTTPS traffic. Tested end to end with mitmproxy 12, certificate checks on: HTTP and HTTPS through mitmproxy and the rotating server, every request out through a US proxy, and a site with an expired certificate refused. Only test what you're allowed to.
+
 **In a pipe** – `-o -` prints the hits to stdout, the interface moves to stderr
 
 ```bash
@@ -834,7 +864,7 @@ curl.exe -x (Get-Content "$run\all.txt" -TotalCount 1) http://api.ipify.org
 | `--serve-refill HOURS` | while serving, check fresh proxies every HOURS and add the hits to the pool |
 | `--serve [PORT]` | afterwards serve as a rotating proxy on `127.0.0.1:PORT` (default: 8899) |
 | `-o FILE` | also write all hits to this file; `-o -` prints them to stdout |
-| `--export FORMATS` | extra files for other tools: `proxychains`, `clash`, `singbox`, `curl` or `all` |
+| `--export FORMATS` | extra files for other tools: `proxychains`, `clash`, `singbox`, `curl`, `pac` or `all` |
 | `-V`, `--version` | print the version |
 | `--completion SHELL` | print the tab completion script for bash, zsh, fish or PowerShell |
 
@@ -920,7 +950,7 @@ Only for things that don't matter. Public proxies are run by strangers who can r
 
 What's next is in the [open issues](../../issues) – ideas and wishes are welcome as an [issue](../../issues/new/choose). During Hacktoberfest there are [beginner-friendly issues](../../issues?q=is%3Aopen+label%3Ahacktoberfest) with pointers on where to start.
 
-- [ ] [Mirror the daily snapshots to a Hugging Face dataset](../../issues/188)
+Shipped in v1.22 and v1.23: a live dashboard for the rotating server, `--export pac` and an hourly `proxy.pac`, "best first" that counts how likely a proxy is still up, the lists in a repository of their own ([free-proxy-list](https://github.com/maximilianfeix/free-proxy-list)), daily snapshots on Hugging Face, `brew install`, a Chinese README, Python 3.14, and a recipe for Burp Suite, ZAP and mitmproxy.
 
 Shipped in v1.10 to v1.21: download speed per proxy and "best first" everywhere, daily snapshots, a rotating proxy URL for requests, httpx and Playwright, `--pick` for working proxies without a scan, a proxy pool API compatible with jhao104/proxy_pool, your own lists with `--source`, a GitHub Action, a Telegram bot and an Atom feed for the weekly report.
 
@@ -969,12 +999,13 @@ proxyscraper/
 ├── geodb.py            DB-IP country database (monthly, binary search)
 ├── targets.py          target sites for --target
 ├── output.py           result files
-├── exporters.py        proxychains, Clash, sing-box and curl formats (--export)
+├── exporters.py        proxychains, Clash, sing-box, curl and PAC formats (--export)
 ├── server/             rotating proxy server (--serve): pool · http · upstream · socks · status · api · core
 ├── api.py              find_proxies() / check_proxies() for Python
 ├── agent.py            MCP tools without the SDK: live list, filters, fetch through proxies
 ├── mcp_server.py       MCP server (proxy-scraper-mcp) for AI agents
 ├── publish.py          live list for GitHub Actions
+├── mirror.py           the same lists in their own repo (free-proxy-list)
 ├── paths.py            where state and results are stored
 ├── compat.py           differences between Unix and Windows
 ├── netio.py            small HTTP client on asyncio

{proxy_scraper_cli-1.22.0 → proxy_scraper_cli-1.23.0}/proxy_scraper_cli.egg-info/SOURCES.txt

@@ -29,6 +29,7 @@ proxyscraper/history.py
 proxyscraper/judges.py
 proxyscraper/mcp_entry.py
 proxyscraper/mcp_server.py
+proxyscraper/mirror.py
 proxyscraper/netio.py
 proxyscraper/options.py
 proxyscraper/output.py
@@ -50,6 +51,7 @@ proxyscraper/targets.py
 proxyscraper/server/__init__.py
 proxyscraper/server/api.py
 proxyscraper/server/core.py
+proxyscraper/server/dashboard.html
 proxyscraper/server/http.py
 proxyscraper/server/pool.py
 proxyscraper/server/socks.py
@@ -87,9 +89,11 @@ tests/test_learning.py
 tests/test_live_api.py
 tests/test_mcp_server.py
 tests/test_mine_sources.py
+tests/test_mirror.py
 tests/test_options.py
 tests/test_output_stdout.py
 tests/test_own_sources.py
+tests/test_pac.py
 tests/test_packaging.py
 tests/test_pages.py
 tests/test_parsing.py
@@ -105,6 +109,7 @@ tests/test_report.py
 tests/test_robustness.py
 tests/test_rotator.py
 tests/test_serve_api.py
+tests/test_serve_dashboard.py
 tests/test_serve_password.py
 tests/test_serve_refill.py
 tests/test_serve_v2.py
@@ -115,6 +120,7 @@ tests/test_sites.py
 tests/test_sources_json.py
 tests/test_speed.py
 tests/test_speed_ranking.py
+tests/test_survival_ranking.py
 tests/test_tamper.py
 tests/test_targets.py
 tests/test_ui.py

{proxy_scraper_cli-1.22.0 → proxy_scraper_cli-1.23.0}/proxy_scraper_cli.egg-info/requires.txt

@@ -4,7 +4,7 @@ certifi
 [dev]
 pytest
 hypothesis
-ruff==0.16.8
+ruff==0.16.9
 
 [fast]
 

{proxy_scraper_cli-1.22.0 → proxy_scraper_cli-1.23.0}/proxyscraper/__init__.py

@@ -1,7 +1,7 @@
 """proxy-scraper: collects public HTTP/SOCKS4/SOCKS5 proxies from hundreds of sources, checks them
 in parallel with its own protocol handshakes and learns which sources deliver good proxies."""
 
-__version__ = "1.22.0"
+__version__ = "1.23.0"
 
 
 def __getattr__(name):

{proxy_scraper_cli-1.22.0 → proxy_scraper_cli-1.23.0}/proxyscraper/agent.py

@@ -181,8 +181,8 @@ def select(rows: Iterable[dict], protocol: str = "any", countries: Iterable[str]
            elite_only: bool = False, exclude_datacenter: bool = False, exclude_blocklisted: bool = False,
            stable_only: bool = False, max_latency_ms: int = 0, run_hours: int = 1, min_uptime: int = 0,
            works_on: Iterable[str] = (), min_speed_kbps: int = 0) -> List[dict]:
-    """The rows that pass every filter, best first (answer plus download speed) – the same filters as on the
-    website."""
+    """The rows that pass every filter, best first (answer plus download speed, and how likely it's still up) – the
+    same filters as on the website."""
     return best_first(matching(rows, protocol, countries, https_only, elite_only, exclude_datacenter,
                                exclude_blocklisted, stable_only, max_latency_ms, run_hours, min_uptime, works_on,
                                min_speed_kbps))

{proxy_scraper_cli-1.22.0 → proxy_scraper_cli-1.23.0}/proxyscraper/api.py

@@ -208,9 +208,9 @@ async def live_proxies_async(*, types: Iterable[str] = PROXY_TYPES, countries: I
                              anonymity: str = "", max_latency: int = 0, no_datacenter: bool = False,
                              no_blocklisted: bool = False, min_uptime: int = 0, works_on: Iterable[str] = (),
                              min_speed: int = 0, limit: int = 0) -> List[LiveProxy]:
-    """The proxies from the hourly list that pass the filters, fastest first – by first answer plus measured
-    download speed. Nothing is checked here: they worked from GitHub's servers in the last run (at most an hour
-    ago). find_proxies checks from your network.
+    """The proxies from the hourly list that pass the filters, best first – by first answer plus measured
+    download speed, and how likely each one is still up. Nothing is checked here: they worked from GitHub's servers
+    in the last run (at most an hour ago). find_proxies checks from your network.
 
     Same filters as find_proxies, plus
     min_uptime  only proxies listed in at least this share (percent) of the week's runs, e.g. 90
@@ -220,7 +220,7 @@ async def live_proxies_async(*, types: Iterable[str] = PROXY_TYPES, countries: I
     limit       at most this many (0 = all)
     """
     from .agent import AgentError, LiveSource
-    from .ranking import page_ms, typical_speed
+    from .ranking import expected_ms, typical_speed
     from .sites import SITE
 
     types = _types(types)
@@ -245,9 +245,11 @@ async def live_proxies_async(*, types: Iterable[str] = PROXY_TYPES, countries: I
         raise ConnectionError(str(e)) from None
     found = [_live_proxy(r, data.run_hours) for r in data.rows
              if r.get("ptype") in types and isinstance(r.get("proxy"), str)]
-    # fastest first – by how quickly a page comes through when the list has download speeds (#186)
+    # best first – by how quickly a page comes through when the list has download speeds (#186), and how likely the
+    # proxy is still up
     typical = typical_speed(p.speed_kbps for p in found)
-    found.sort(key=lambda p: page_ms(p.latency, p.speed_kbps, p.https, typical))
+    run_hours = data.run_hours or 1
+    found.sort(key=lambda p: expected_ms(p.latency, p.speed_kbps, p.https, typical, p.up_for_hours // run_hours))
     found = [p for p in found if opts.filters.accepts(p) and (not min_uptime or (p.uptime_7d or 0) >= min_uptime)
              and all(p.sites.get(s) for s in works_on) and (not min_speed or (p.speed_kbps or 0) >= min_speed)]
     return found[:limit] if limit else found