The extension fetches model lists and sends reload commands to the user's own KoboldCPP server, which is exactly what it advertises. The 'code execution' the scanner noticed is just programmatically clicking SillyTavern UI buttons to reconnect after a model reload — no hidden or dangerous code is run.
Technical evidence
Scanner reason: JavaScript analysis matched static JavaScript security signal javascript.download-to-execution. The match applies to this repository.
Contextual assessment: The scanner correlates network fetch primitives (lines 53 and 73) with what it detects as a code execution sink. Examining the actual code, the fetch calls go to user-configured KoboldCPP admin API endpoints (list_options and reload_config) — the stated purpose of the extension. The suspected execution sink is jQuery .click() calls on lines 86 and 89, which simulate UI button presses to trigger SillyTavern's API reconnection flow. These are not eval, Function constructor, or any dynamic code execution primitive. There is no credential exfiltration, no external destination beyond the user-specified KoboldCPP URL, no obfuscation, and no persistence beyond standard extension_settings. All data flow is consistent with the project's README-described purpose of managing KoboldCPP model loading.