Skip to content

Zero Breach — local, read-only scanner that checks a computer for indicators of the AdFection / Candiru ad-delivered spyware campaign. No network, no admin, no dependencies.

Clarify documented scan coverage and offline limits

interplainsefbe9079d887
diff --git a/CONTRIBUTING.md b/CONTRIBUTING.md
index 66c2a1a..30ae9ab 100644
--- a/CONTRIBUTING.md
+++ b/CONTRIBUTING.md
@@ -49,13 +49,16 @@ person to a crisis helpline. So:
These are enforced by tests, not convention. A change that breaks one will fail CI.
-- **Never contact an indicator.** The scan runs inside a guard that makes any
- network access raise. The one network feature, `update-iocs.py`, is a separate
- script, never invoked during a scan.
+- **Never contact an indicator.** The scan's Python socket guard blocks direct
+ DNS and network calls. It is not an OS network sandbox: POSIX network mounts
+ and concurrent filesystem changes require OS isolation for strict offline use.
+ The network feature, `update-iocs.py`, is a separate script, never invoked
+ during a scan.
- **Measured, never inferred.** A check that could not run is a reported gap, never
a clean result.
-- **Read only what is needed.** Cookie host names, never cookie values; never the
- credential-decryption paths an infostealer uses.
+- **Query only what is needed.** Cookie queries select host metadata and
+ timestamps, not values. Temporary snapshots copy complete database bytes;
+ the scanner does not decrypt browser credentials.
- **Precision over coverage.** An unsigned binary in a user-writable path is not a
finding. Alarms that mean nothing teach people to ignore the tool.
- **Read-only, no administrator, no telemetry, no third-party dependencies,
diff --git a/README.md b/README.md
index dc48955..6f23055 100644
--- a/README.md
+++ b/README.md
@@ -43,15 +43,16 @@ updates or establish that a device is safe.
| area | what it reads |
|---|---|
-| **Browser** | history, cookies (host names and timestamps only), favicons, HTTP cache, Local/Session Storage, Service Worker caches, and HSTS/connection records — across Chrome, Edge, Brave, Chromium and Firefox, every profile |
+| **Browser** | history, cookies (host names and timestamps only), favicons, HTTP cache, Local/Session Storage, Service Worker caches, and HSTS/connection records — across Chrome, Edge, Brave, Chromium and Firefox, for discovered profiles in supported locations |
| **Files** | SHA-256 of candidate files under the user-writable locations spyware drops into, against known sample hashes — and, kept deliberately separate, the hash of a legitimate signed driver this actor abuses, which is reported as a lead rather than as proof |
| **Host** (Windows) | per-user COM registrations (this campaign's documented persistence), the DNS resolver cache, whether DNS history logging is even enabled, the hosts file, and known fixed-path driver drops (e.g. the `HW.sys` vulnerable driver) |
-Every match is **evidence-graded**, because the artifact decides the meaning. A
-cookie the browser set, or an HSTS record it stored, proves the device *contacted*
-a host. A domain merely appearing in a cached web page usually means the person
-*read an article about* the campaign. The tool keeps those apart so that reading
-about this topic never produces a false alarm, and a real contact is never buried.
+Every match is **evidence-graded**, because the artifact decides the meaning.
+Cookie metadata and stored connection records can support a contact finding,
+but do not establish infection or explain how the record arrived. A domain
+merely appearing in a cached web page may be a saved reference, such as an
+article about the campaign. The tool distinguishes saved references from
+connection evidence so that each finding can be interpreted in context.
## Install and run
@@ -74,7 +75,8 @@ python -X utf8 scan/adint-scan.py
The scan changes no system settings and needs no administrator. It makes no
intentional network requests or indicator lookups. Remote and redirected paths
are skipped, and online certificate verification is disabled. Locked database
-snapshots are temporary and removed after use; `--json` writes the requested report.
+snapshots are temporary; cleanup is attempted after use, and any cleanup failure
+is reported with the retained location. `--json` writes the requested report.
Exit codes: `0` nothing found, `2` an indicator matched, `1` the scan could not be
set up (e.g. a bad indicator file).
@@ -219,7 +221,7 @@ of spyware.
The list goes stale — new infrastructure appears and old domains lapse. Refresh
it from this repository with a separate, explicitly-run script (the scanner itself
-never touches the network):
+makes no intentional network requests):
```
python scan/update-iocs.py # fetch, verify, show what changed (dry run)
@@ -256,8 +258,9 @@ cd scan && python -m unittest discover -s tests
The suite covers the network guard, the evidence grading, the boundary matching
(so a longer name or number that merely contains an indicator does not match), the
-loader's failure modes, and behavioural proofs that the tool never reads cookie
-values or the browser's master key.
+loader's failure modes, and behavioural checks that cookie queries exclude values
+and that the scanner does not read the browser's master key. Snapshot copies still
+contain the database bytes, as described above.
## Contact
diff --git a/SECURITY.md b/SECURITY.md
index a1b7c29..5790d64 100644
--- a/SECURITY.md
+++ b/SECURITY.md
@@ -46,7 +46,8 @@ source and confirm it matches. See "Verifying what you downloaded" in the README
## Handling indicators responsibly
The indicator set is curated to avoid false alarms against real people — it
-excludes compromised victim sites and domains that could belong to a real
-organisation. If you believe an indicator here is wrong, or would match legitimate
-traffic, that is worth reporting: a false positive from this tool can send an
+excludes known compromised victim sites and ambiguous legitimate domains during
+curation. Historical domains can later change ownership. If you believe an
+indicator here is wrong, or would match legitimate traffic, that is worth
+reporting: a false positive from this tool can send an
innocent person to a crisis helpline.