De-Noise: a HestiaCP plugin to reduce bot traffic before it reaches PHP

Hi everyone,

After migrating several websites to HestiaCP, I was seeing plenty of unnecessary traffic: requests for exposed credentials, nonexistent PHP files, aggressive crawlers and repeated WordPress login attempts.

So I put together the De-Noise to manage this centrally through the Hestia panel, filtering requests in Nginx before they reach Varnish, Apache or PHP.

Some of its features:

  • Separate, editable scanner-tool and crawler blocklists.
  • Protection against known private file and credential probes.
  • Optional login, xml-rpc and filtered URL rate limits.
  • Optional automatic IP bans through Hestia’s firewall, plus domain only bans.
  • Ban explanations showing the requests that triggered them.
  • WordPress user enumeration protection and Fail2ban login controls.
  • Extra HTTP password for WordPress administration
  • Country filtering using a local GeoLite2 database.
  • Central access defaults with account and domain overrides.
  • Log analysis and Activity reports with site filters and full User agent details.

Existing proxy templates remain. Each domain supports Off, Observe and Enforce modes.

This is an independent, open-source project, currently preview version 0.16.0, targeting Ubuntu with Hestia’s Nginx proxy setup. It is a request-filtering addon, not a full WAF.

Repository, installation instructions and documentation:
github.com/orloxgr/hestia-de-noise

Feedback would be especially useful around false positives, custom templates and legitimate WordPress/WooCommerce workflows. If you try it and you have an issue, please include your setup and the specific rule involved when reporting an issue.
Suggestions are also more than welcome.

  • Can it be used on debian?
  • Is this only meant for wordpress?

The current release has been tested on Ubuntu 26, and the installer currently only accepts Ubuntu.
Deleting from /usr/local/src/hestia-de-noise/tools/install.py these lines:

os_release = Path(‘/etc/os-release’).read_text()
if not any(line in os_release.splitlines() for line in (‘ID=ubuntu’, ‘ID=“ubuntu”’)):
raise ValueError(‘This release targets Ubuntu with HestiaCP’)

you should be able to install it on debian but I cannot know if it will work or even if it will break your sytem or not :stuck_out_tongue:

The general features work at the Nginx level and can protect PHP applications, static HTML sites and other websites: bot filtering, private file protection, rate limits, optional IP bans, country filtering and log analysis.

The Wordpress specific features such as user enumeration protection, failed login integration and WordPress cron management are optional. You don’t need Wordpress to use the general filtering features.

Extremely useful tool! Will provide feedback after testing. Thank you!

Is it possible to protect a specific url(like /wp-admin) serverwide with captcha(something as simple as a math/image captcha) protection?

it is! and now it does! you can choose between a password and the captcha.

Installed it without issues on a test u26.06 vps.

Instead of just showing json output, its easy if you show a success message for novices.

python3 /usr/local/src/hestia-de-noise/tools/install.py
{"ok": true, "version": "0.18.2", "mode": "observe", "panel": "/plugins/de-noise/", "changed_files": 3, "sites": 2, "supported_sites": 2, "wordpress": {"managed": false}, "unsupported": }

you are right. I prettyfied it a bit with install progress.

Like User-Agent lists to block, it will be useful if we have custom (regex enabled)list of endpoints to block.

For example, i often see bad scan bots looking for bak.sql.gz, bak.sql.zip, bak.sql.zst etc. If have the above feature, we can block and eliminate much noise using a regex.

You can already do this in IP bans. “Ban IPs that 404 any of these files X times”.

You can analyze php 404s and put the bak.sql.gz, bak.sql.zip, bak.sql.zst in the list.

edit: Ignore the above, now i understand what you mean and it is also a great idea. Ill work on this.

edit2: it is now implemented

added an option to disable xml-prc on websites not needed.

Nice tool :slight_smile: I’ve been blocking bots at the nginx level. Here are the one we use that you may want to add to the suggested

      ~*Amazonbot                   1;

        ~*Amazon\ Web\ Services       1;

        ~*DataForSeoBot               1;

        ~*Bytespider                  1;

        ~*SemrushBot                  1;

        ~*DotBot                      1;

        ~*MJ12bot                     1;

        ~*AwarioBot                   1;

        ~*ImagesiftBot                1;

        ~*ClaudeBot                   1;

        ~*GPTBot                      1;

        ~*Barkrowler                  1;

        ~*Owler                       1;

        ~*GoogleOther                 1;

        ~*SeekportBot                 1;

        ~*serpstatbot                 1;

        ~*BLEXBot                     1;

        ~*YandexBot                   1;

        ~*OAI-SearchBot               1;

        ~*BacklinksExtendedBot        1;

        ~*VelenPublicWebCrawler       1;

        ~*AhrefsBot                   1;

        ~*CCBot                       1;

        ~*Timpibot                    1;

        ~*PetalBot                    1;

        ~*ZoominfoBot                 1;

        ~*MojeekBot                   1;

        ~*SearchAtlasBot              1;

        ~*LinkpadBot                  1;

        ~*Magpie-Crawler              1;

        ~*Screaming\ Frog\ SEO\ Spider 1;

        ~*IbouBot                     1;

        ~*360Spider                1;

        ~*Go-http-client             1;

        ~*SERankingBacklinksBot      1;

        ~*RecordedFuture              1;

        ~*PerplexityBot               1;

        ~*BaiduSpider                 1;

        ~*MegaIndex\.ru               1;

        ~*SeobilityBot                1;

        ~*SEOkicks-Robot              1;

        ~*rogerbot                    1;

        ~*quic-go                     1;

        ~*RecordedFuture              1;

        ~*CMS-Checker                 1;

        ~*Claude-SearchBot            1;

        ~*ExaSearchBot                1;

        ~*Reflectionbot                  1;

        ~*AionBot                     1;

        ~*Epochal-Search                1;

        ~*BasegrowBot                  1;

nice, thank you for the list. Now denoise suggest these into their category. Added a list for major and various search engine bots since its not easy to distinguish what some may want or not.