The Agent Skills Directory

[COMMAND_EXECUTION]: The administrative script scripts/backfill_pdf_metadata.mjs utilizes node:child_process.execFile to invoke the pdftotext and pdfinfo utilities. These operations are designed for manual execution by a trusted operator to process local PDF files and are not invoked automatically by the AI agent during typical interactions.
[EXTERNAL_DOWNLOADS]: The skill and its associated scripts communicate with external Busabase API endpoints. The connection details, including the target URL and API credentials, are managed by the operator through environment variables (BUSABASE_URL, BUSABASE_API_KEY).
[INDIRECT_PROMPT_INJECTION]: The skill processes and displays insurance records and text extracted from PDF documents, which serves as a potential attack surface for indirect prompt injection. Ingestion points: Data is fetched from Busabase Drive nodes and Base records, and text is extracted from local PDF files. Boundary markers: The application implements HTML escaping (escapeHtml) for all data rendered in the UI. Capability inventory: The skill includes network capabilities via fetch and local command execution via execFile in operator scripts. Sanitization: UI content is sanitized using escaping mechanisms before interpolation.
[DYNAMIC_EXECUTION]: The development script scripts/sync-content.mjs uses dynamic import() to load the application configuration. This process is used for synchronizing content sidecars with the main configuration and does not involve the execution of untrusted code at runtime.

kelly-insure-data