The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses Node.js scripts (plan-component-update.mjs, build-importable-component.mjs, and validate-importable-component.mjs) to process JSON specifications and generate artifacts. These operations are local, restricted to a declared staging directory, and follow the principle of least privilege.
[DATA_EXFILTRATION]: No network operations or external data transmissions were detected. The skill operates entirely on local files provided by the user or generated during the build process.
[PATH_TRAVERSAL_PREVENTION]: The build-importable-component.mjs script explicitly validates that generated output paths do not escape the staging root using path.resolve and prefix checks (!target.startsWith(...)).
[ATOMIC_OPERATIONS]: The build script implements an 'exclusive reservation' pattern using the wx flag in fs.openSync to ensure it does not overwrite existing files. It also features a manual rollback mechanism that attempts to delete partially written files if a batch operation fails.

sillytavern-component-update