Skip to main content
← Tools
Tool · Code repo

Check for Spyware Traces Without Uploading Your Data

Check for traces of ad-delivered AdFection / Candiru spyware on your own computer. Zero Breach reads local browser records, file fingerprints and supported Windows records, then reports the evidence and checks it could not complete. Scan results stay on your computer.

@interplains Zero Breach
  • Spyware investigation
  • Spyware
  • Forensics
  • Privacy
  • Incident response

Requirements

  • Python >= 3.8
  • Windows / macOS / Linux

Overview

What you get

  • Local checks against published AdFection / Candiru indicators, without uploading scan results or requesting administrator access.
  • Findings that distinguish saved references, connection records, file matches and persistence evidence, with the supporting locations.
  • A readable summary, a list of checks that could not run, and an optional JSON report saved to a local path.

Good fit

  • Checking surviving records on your own computer for known indicators of this spyware campaign.
  • Preserving findings and coverage details for further investigation.

Tutorial

Understand the checks and their results

To investigate suspicious activity on your computer, a count of matched domains is not enough. You need to know where each match came from: saved page text, a connection record or a file on disk.

Zero Breach checks published AdFection / Candiru indicators for traces of spyware delivered through advertising. It runs locally, keeps results on your computer and does not change system settings.

AreaWhat it reads
BrowsersHistory, cookie host names and timestamps, caches, storage and connection records in common profile locations for Chrome, Edge, Brave, Vivaldi, Chromium and Firefox
FilesSHA-256 fingerprints of candidate files in common user-writable locations, compared with known sample and related driver fingerprints
WindowsPer-user COM registrations, DNS cache and logging status, hosts entries and known fixed file paths

Start with a quick check, then include file fingerprints and save a report. Read the summary, evidence locations and coverage gaps together.

Download and verify the single-file tool

This walkthrough uses Windows PowerShell and Python 3.8 or newer. On macOS/Linux, replace python with python3 in the scan commands below; Windows host checks run only on Windows. Default checks need no extra Python packages or administrator access.

  1. From the v0.1.8 release, download zero-breach.py, zero-breach.py.asc and archivegenocide-signing-key.asc into one local folder. Open PowerShell in that folder.
  2. Make sure GnuPG’s gpg command is available. Windows users can follow the Gpg4win link on the publisher’s verification page. Once gpg is available, display the downloaded key’s fingerprint:
gpg --show-keys --with-fingerprint archivegenocide-signing-key.asc

Compare the full fingerprint with the publisher’s verification page: C24E C92B 12D6 670A 2516 065F 9B4D 5754 99AF A53C. After confirming the match, import the key and verify the file:

gpg --import archivegenocide-signing-key.asc
gpg --verify zero-breach.py.asc zero-breach.py

These two lines are from our verification of v0.1.8:

gpg:                using EDDSA key C24EC92B12D6670A2516065F9B4D575499AFA53C
gpg: Good signature from "Archive Genocide <archive@archivegenocide.com>" [unknown]

Good signature confirms that the signature is valid. Also check that the full fingerprint after using EDDSA key matches the one above. See SIGNING.md for the full procedure.

  1. Check the downloaded file’s SHA-256:
Get-FileHash zero-breach.py -Algorithm SHA256

For v0.1.8, expect 74740af20f609f82dc6e50d66f234d0c39a20f516caa35731c2d19c11684ef44. Letter case does not affect the comparison. On macOS/Linux, use shasum -a 256 zero-breach.py.

  1. Confirm your Python version, then check the tool’s version:
python --version
python -X utf8 zero-breach.py --version

The tool should print Zero Breach 0.1.8 (single-file build). Continue with the scan once the signature is valid and the fingerprint, version and checksum all match.

Run a quick check first

  1. Close the browsers you want to check so their databases are no longer being updated or held open.
  2. Stay in the download folder and run this in PowerShell:
python -X utf8 zero-breach.py --quick
  1. When it finishes, read SUMMARY, then WHAT COULD NOT BE CHECKED, then the individual findings and their locations.

--quick skips file fingerprints and reports that omission as a gap. If Windows DNS history logging is disabled, the report also explains that no history is available. Rerunning cannot recover events that were never logged.

If there are no matches, check which browser profiles and records the tool actually found. Opera, Arc and custom profile locations are outside its current support; existing caches and history cover only records that remain on the device.

Include file checks and save a local report

Leave out --quick to include file fingerprints in the default checks. Add --json to save the report in the current folder:

python -X utf8 zero-breach.py --json report.json

The terminal still shows the readable report and prints the JSON output path. Confirm that report.json exists in the current folder, open it and check three areas:

FieldWhat to confirm
timestamp, platformThe report belongs to this run and system
modulesEach check’s status, findings and evidence locations
verdict.gapsChecks that did not complete and their reasons

Compare it with the quick check: file fingerprinting should now have run, or the report should explain why it could not. Read any remaining gaps individually.

Exit code 0 means no indicators matched, 2 means an indicator matched, and 1 means the scan could not initialise. For code 2, continue with the report’s evidence; it does not mean the program crashed.

Choose the next step from the evidence

For each finding, consider the evidence type, indicator role, time and location. Start with these categories:

Report labelHow to interpret it and continue
referencedA domain appears in saved page or storage content. Read the context; it may be an article about the campaign
dns-recordA domain or address appears in local DNS records. Resolution alone does not establish a connection
contactedThere is a connection-related record. Check whether the indicator belongs to advertising delivery, command-and-control or other infrastructure linked to the campaign
configuredLocal configuration maps the domain to a real address. Inspect the configuration and its purpose
on-diskA file fingerprint matched. Distinguish sample (a known spyware sample) from vuln-driver (a legitimate driver the actor abuses)
persistenceAn auto-start registration is associated with an indicator. Preserve its full location for investigation

For a known-sample or persistence finding, preserve the report and evidence paths, then contact a qualified team such as the Access Now Digital Security Helpline. For saved references alone, inspect the content first. For no matches, use the coverage information to decide what remains to check.

To add pattern checks for selected files, read the official optional-check guide. This mode needs Python 3.9+ and the pinned YARA-X dependency. --pattern-path limits only that extra check; browser, fingerprint and host checks keep their usual scope.