The extension downloads and saves small fact lists using the chat app's own built-in save/load features, the same way the app stores your other files. What it downloads is plain text facts that only get added to the AI's prompt. Nothing it fetches from the network is ever run as code or a command, so the scanner's warning about 'download then execute' does not match what actually happens.
Technical evidence
Scanner reason: JavaScript analysis matched static JavaScript security signal javascript.download-to-execution. The match applies to this repository.
Contextual assessment: The flagged network retrieval consists of fetch calls to the SillyTavern host's own endpoints: a GET of a per-character facts file under the server's user files route and a POST to the server's file-upload API, both authenticated with the host-provided request headers and credentials. The retrieved data is JSON containing fact strings that are trimmed, validated, and only ever rendered into prompt text via template substitution; nothing is parsed as code, evaluated, injected as markup, or passed to a command interpreter. The scanner's 'execution sink' correlation appears to be a keyword-level match over the same file, not an actual data flow from the network response into a dynamic execution or command path. This behavior matches the project's stated purpose of persisting and injecting per-character user facts, and all destinations are the user's own SillyTavern server.