The Agent Skills Directory

[SAFE]: The skill demonstrates a strong security posture by managing credentials through environment variables rather than hardcoding. It uses authorized Busabase SDK methods for data operations and maintains a clear boundary between UI actions and backend executions.
[INDIRECT_PROMPT_INJECTION]: The skill handles external product data (exports, CSVs), which creates a standard attack surface for indirect prompt injection. This risk is minimized by the skill's implementation of data escaping and its operational design.
Ingestion points: The scripts/ingest_products.mjs script acts as the entry point for product, inventory, and channel data from external files.
Boundary markers: While SKILL.md lacks explicit structural delimiters for untrusted data, it provides strict behavioral constraints for the agent regarding data validity and evidence requirements.
Capability inventory: The skill is limited to reading/writing its own provisioned Busabase resources (products, channels, inventory, review, settings) and does not possess broad file system or network permissions beyond its defined scope.
Sanitization: Data displayed in the AirApp is sanitized through dedicated escaping functions like esc() in app.js and escapeHtml() in accent-theme.js, preventing stored XSS or related injection issues.

kelly-products