kelly-listing
Pass
Audited by Gen Agent Trust Hub on Aug 25, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it ingests untrusted product source material to generate e-commerce listings.
- Ingestion points: Product specifications, features, and kelly-picks handoff briefs are ingested into the Busabase backend via the
scripts/ingest_drafts.mjsscript. - Boundary markers: The skill relies on JSON structure and explicit field validation defined in
listing-model.jsto separate data from instructions, but does not implement explicit delimiters or LLM-level 'ignore' warnings for the drafted content itself. - Capability inventory: The skill possesses capabilities to write to the local filesystem (via
scripts/export_listings.mjs) and perform network operations to a user-configured Busabase instance. - Sanitization: Input is validated for type and shape in
scripts/ingest_drafts.mjs, and field values are normalized and slugified to prevent file system traversal during export.
Audit Metadata