Fill in the Blank Generator
Based on the RiskyPlugins AI security review of the observed evidence. The review covers version 1.0.0, not the analysed version 1.0.5.
Analysis record
- Analysed
- Yesterday 20 November 2026
- Version
- v1.0.5
- Artifact
- SHA256 3470473a…a55f0b
- Detail
- Full list
Is Fill in the Blank Generator safe?
Fill in the Blank Generator turns a passage or a photo into a gap-fill worksheet with an answer key, aimed at ESL teachers. It asks for no special permissions: the permission list and the host permission list are both empty, so it cannot read the pages you browse or reach your accounts. The only network activity in the code is a small set of fetch calls, logged as NET-FETCH-background.js-49 and NET-FETCH-panel.js-2525, with no destination domain recorded anywhere in the extension.
The findings that look alarming on a chart are three hits for hidden zero-width characters, listed as OBFUSCATION-INVISIBLE_ZERO_WIDTH in _locales/fa/messages.json, _locales/kn/messages.json and _locales/te/messages.json. Those are the Farsi, Kannada and Telugu translation files, and the characters the scanner found are ordinary parts of those alphabets. The hundreds of "domains" in the endpoint list, things like aacute.sc and aogonek.sngstory.retrohook.sc, are font glyph names that a parser mistook for web addresses.
What the scanner tripped on is the shape of minified JavaScript and the Unicode inside translated strings. Neither is the extension misbehaving. Nothing in the code matched a malware signature, and the extension never names a server it would send your text to. If you are deciding whether to try it, the evidence shows no harmful capability, only scanner noise.
Evidence ledger
Ranked by severity · findings with a source location link to the code viewer
Publisher Evidence
Limited evidencedev--2d80c111-b010-2c14-7257-8d56736a332b
Publisher identity, store signals, distribution reach, and warning signals used for context. Treat this as supporting evidence, not a clean bill of health.
Chrome does not expose the same publisher verification data as IDE stores, so this score is deliberately conservative.
12 evidence rows available.
Finding Categories
Requested Permissions
4 permissionsAI Security Report
AI Security Review
Evidence context: threat category none; evidence quality moderate.
"Fill in the Blank Generator" is a Chrome extension that turns a passage or a photo into a gap-fill worksheet with an answer key. Version 1.0.0, published under [email removed], zero recorded installs.
The manifest asks for nothing. Both the permission list and the host permission list are empty, so the extension has no declared ability to read the pages you visit, touch cookies, or send browsing data anywhere. The only network activity in the bundle is eight generic fetch calls: NET-FETCH-background.js-49, -50 and -51 in background.js, NET-FETCH-panel.js-2525, -2526, -2527 and NET-FETCH-panel.js-3012 in panel.js, and NET-FETCH-pdfout.js-476 in pdfout.js. Those titles record that a fetch exists at those lines and nothing more. No destination is named, and no external service domain appears anywhere in the extension.
The 828 IoC entries explain why the raw count looks alarming, and they are the loudest noise here. The endpoint list reads a0.dataset.lv, aacute.sc, abreve.sc, aogonek.sngstory.retrohook.sc, acircumflex.sc. Those are OpenType glyph names: "sc" for small caps, "sngstory" and "retrohook" for stylistic alternates. The extractor has parsed font metadata or CSS feature strings as though they were hostnames. A real exfiltration endpoint would be one recurring domain; these run in strict alphabetical Unicode order from a to z.
The three high-severity hits are all OBFUSCATION-INVISIBLE_ZERO_WIDTH in _locales/fa/messages.json, _locales/kn/messages.json and _locales/te/messages.json. Those are the Farsi, Kannada and Telugu translation files. Zero-width joiners and non-joiners are required characters in those writing systems, and any scanner hunting hidden Unicode will flag them every time. The remaining 25 findings are low-severity code-smell rules of the kind that fire on any modern JavaScript bundle. Malware signatures: zero. Tool-poisoning: zero. Secrets: zero.
A skeptic would point at the store description promising generation "from any text or photo", which implies OCR or a language model behind an API, and note that background.js and panel.js do contain fetch calls despite no host permission being declared. If the extension reaches a remote service, where is the permission? The straightforward reading is that those fetches load files bundled inside the package or assemble blobs for the PDF path in pdfout.js, which is exactly what a worksheet exporter needs. If it did contact an OCR or AI backend, that hostname would surface in the scan; instead the only "domains" are glyph names. That is the evidence that would change this verdict: a single recurring API hostname in background.js paired with a host permission for it. Nothing here matches that shape.
Every high-severity finding traces to a known scanner pattern rather than to code that does something to a user.
Key Reasons
- All three high-severity findings are OBFUSCATION-INVISIBLE_ZERO_WIDTH hits in _locales/fa, _locales/kn and _locales/te message files, the documented locale false positive
- Zero malware signatures, zero tool-poisoning findings and zero secrets across the entire bundle
- The 828 IoCs resolve to OpenType glyph names such as aacute.sc and aogonek.sngstory.retrohook.sc rather than real hostnames
- The extension declares no permissions and no host permissions
- The network findings are undirected fetch calls in background.js, panel.js and pdfout.js with no external endpoint named anywhere
False Positive Considerations
- Zero-width Unicode characters in Farsi, Kannada and Telugu locale translation files
- IoC extractor parsing OpenType glyph names (small caps, stylistic alternates) as network domains
- Generic fetch calls in background.js, panel.js and pdfout.js flagged without a destination endpoint
- Low-severity code-smell rules firing on bundled JavaScript
Reviewed 2026-10-02 · version 1.0.0; recommended action: suppress false positive; model confidence 85%.
Chrome version history
Risk trend by version
4 analyzed versions. Each point is the latest successful scan for that version; failed zero-score scans are hidden. Dates: First seen by Risky Plugins.
Pick any point on the chart, then open the Source tab to read that version's code.
Source Code Not Available
Source code is not available for this version of the extension.
About This Extension
Frequently Asked Questions
Similar Extensions
Related extensions from the same publisher or marketplace