The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires the manual execution of administrative scripts (node scripts/setup.mjs, node scripts/generate_batch.mjs, node scripts/execute_decisions.mjs) to provision the Busabase environment and seed synthetic data. These operations are restricted to a local environment and are intended for initial configuration.
[DATA_EXFILTRATION]: The administrative scripts leverage environment variables (BUSABASE_API_KEY, BUSABASE_SPACE_ID, BUSABASE_BASE_URL) for authentication. This follows established best practices for secure credential management by avoiding hardcoded secrets.
[PROMPT_INJECTION]: The skill creates a potential surface for indirect prompt injection by ingesting data from external Busabase bases (vehicles, items).
Ingestion points: Data is read from Busabase bases via busabase-provider.js.
Boundary markers: The skill relies on standard data separation; specific boundary instructions to the agent regarding the content of these bases are not present.
Capability inventory: The skill can write back to Busabase via the submitReview function and shell script execution.
Sanitization: The AirApp implements escapeHtml in app.js to mitigate common injection risks during UI rendering.

kelly-disclosure-tracker