WAPIN

AI Security Review

Evidence context: threat category none; evidence quality moderate.

WAPIN describes itself as a CRM for WhatsApp Web, and its files match that description. The extension ships a background.js plus two content script trees, content/assets/js/ and funnil/assets/js/, and the network findings inside them are the ordinary plumbing of a chat client: NET-SOCKET_IO at content/assets/js/v_7_4_3_86_085ff17f-7684-4a41-bfdd-8bd9d5293bb212.js:15, NET-JQUERY_AJAX at content/assets/js/v_7_4_3_86_085ff17f-7684-4a41-bfdd-8bd9d5293bb223.js:13, and a dozen NET-FETCH hits across the same directory plus background.js:137. Socket.IO and jQuery are the libraries a chat CRM uses to hold a live connection and poll a backend. None of these findings names a destination host, so they describe capability, the ability to talk to a server, rather than any specific data path out.

The endpoint list is where the noise becomes plain. Alongside app.wascript.com, which reads as the vendor's own service, the scanner reports border.radius.circle, color.border.accent.orange, btn.glass, artboard.phone, above.sm, calls.function.arguments.delta and avatar.online. Those are CSS design tokens, utility class names and a streaming API field name lifted out of minified JavaScript, and the extractor matched dot-separated identifier chains as if they were hostnames. api.statsigcdn.com and the Atlassian API hosts carry the same problem one step removed: they arrive from bundled third-party code and embedded strings, not from code that posts chats anywhere.

Obfuscation accounts for two findings. OBFUSCATION-LARGE_BASE64 in funnil/assets/js/ed6113f2-bc09-4421-a453-bbbd66fe9488.DCssQ7AT.js:1 is a base64 blob, which in a webpack-style asset build is usually an inlined font, image or JSON payload. OBFUSCATION-FUNCTION_INDIRECT in content/assets/js/v_7_4_3_86_085ff17f-7684-4a41-bfdd-8bd9d5293bb2259.js:1 matches the call indirection any minifier produces. Neither sits beside a malware match. The scan returned zero malware family signatures of any kind across 1,201 findings, and zero critical or high severity items. What remains is 754 code-smell hits and 422 indicator matches, the documented noise floor of bundled JavaScript.

A skeptic would push back hard here, and reasonably. WAPIN reads WhatsApp Web, so every message, contact and group name a user has open sits within its reach, and it opens a socket to a third-party domain. That is a genuine capability. It is also why we treat this as a powerful extension that asks for trust rather than as a harmless one, and why the destination of that Socket.IO channel sits outside what static reading can settle. What the code does show is the absence of harvesting behaviour: no login-domain indicators, no cookie reads paired with transmission, no file reading, and no malware signature anywhere in the tree. The count is high because 50-odd bundled libraries each trip their own generic rules, which multiplies one benign pattern into hundreds of rows. One limitation is worth stating plainly: the recorded permission and host_permission lists come back empty, so we cannot compare what WAPIN asks for in its manifest against what its scripts do. That gap, not the finding volume, is the thing to check on the next release.

Key Reasons

• Zero malware signatures and zero critical or high severity findings across 1,201 matches in a versioned, identifiable release
• Network findings are library-typical for a chat client: NET-SOCKET_IO, NET-JQUERY_AJAX and NET-FETCH in content/assets/js/ and background.js:137, none naming a destination host
• Reported endpoints include CSS design tokens and code fragments (border.radius.circle, btn.glass, calls.function.arguments.delta, above.sm), showing extractor false positives rather than real hosts
• Obfuscation flags are an inlined base64 asset in funnil/assets/js/ed6113f2-bc09-4421-a453-bbbd66fe9488.DCssQ7AT.js and minifier indirection in content/assets/js/v_7_4_3_86_085ff17f-7684-4a41-bfdd-8bd9d5293bb2259.js, with no malware family co-located
• Publisher is reachable at [email protected] and the extension has only 21 users, a poor hiding place for deliberate malware

False Positive Considerations

• 754 code-smell findings from generic rules matching bundled npm libraries (Node patterns, env vars, eval-style constructs)
• 422 IoC matches that include minified identifier chains and CSS token strings misread as domains
• Obfuscation rules firing on an inlined base64 asset blob and webpack-style function indirection
• High finding count amplified by many bundled libraries each tripping its own YARA and IoC rule

Reviewed 2026-10-10; recommended action: monitor; model confidence 75%.