The scanner flagged text that looks like a web address. It is only the standard identifying label placed inside the data-format descriptions that tell the AI what shape its answer should have. Nothing downloads from those labels; the extension only communicates with the AI service the user already set up in SillyTavern.
Technical evidence
Scanner reason: JavaScript analysis matched static JavaScript security signal javascript.xray.shady-link. The match applies to this repository.
Contextual assessment: The five scanner matches correspond exactly to the five schema-definition constants in this module, each of which carries the standard JSON Schema draft-dialect meta-schema identifier field. These are declarative metadata strings embedded in static objects that describe the expected shape of model output; they are passed to the host application's generation APIs as part of structured-output requests. No network fetch reads them, no credential material is attached to them, and all model traffic routes through SillyTavern's configured connection (active connection or a user-selected connection profile), which matches the project's stated purpose. The flagged literals are not exfiltration destinations and the surrounding module only adds client-side request queuing, pacing, retry classification, and response normalization.
The scanner's obfuscation guess does not match the actual code. What is shown is plainly written interface code for the extension's popup windows and buttons, with no hidden or disguised behavior.
Technical evidence
Scanner reason: JavaScript analysis matched static JavaScript security signal javascript.xray.obfuscated-code. The match applies to this repository.
Contextual assessment: The supplied source is transparent interface logic: construction of the manipulator dialog markup, connection-profile and token-limit controls, review cancellation tokens guarding stale asynchronous results, an in-memory session cache, and list filtering. Lorebook names are HTML-escaped before insertion into markup and numeric settings are clamped integers. No minification, encoding, dynamic evaluation, or hidden control flow appears in the provided window; the low-confidence scanner signal is a heuristic false positive on the long readable template markup and does not correspond to any demonstrated obfuscation. Behavior matches the stated purpose of reviewing, drafting, and applying user-approved lorebook edits with backups.
The automated scanner guessed the file might be deliberately hard to read, but the actual code shown is ordinary, clearly written extension code. Nothing is hidden, encoded, or disguised, and every action it takes fits its advertised purpose.
Technical evidence
Scanner reason: JavaScript analysis matched static JavaScript security signal javascript.xray.obfuscated-code. The match applies to this repository.
Contextual assessment: The supplied source for the extension entry module is fully readable ES-module code: local module imports, extension settings initialization with clamped numeric bounds, event handlers, slash-command registration, backup restore/download flows gated by confirmation dialogs, and UI icon injection via a mutation observer. No encoded blobs, dynamic evaluation, packed payloads, string-concatenated indirection, or any concealment technique appears anywhere in the provided window. The low-confidence scanner signal at line one is not corroborated by any demonstrated obfuscation, and all observed data flow matches the stated lorebook-maintenance purpose with persistence limited to the host application's extension settings.