dynamic-pricing

Pass

Audited by Gen Agent Trust Hub on Mar 16, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill follows security and operational best practices for automated financial decision-making. It implements a layered defense strategy including absolute margin floors (15% over cost), change caps (20%), and mandatory human approval for significant price movements, which prevents both algorithmic errors and external manipulation from causing financial damage.
  • [PROMPT_INJECTION]: The skill processes untrusted external data (competitor prices) which represents a surface for Indirect Prompt Injection.
  • Ingestion points: Scrapes pricing information from external competitor URLs via Playwright as defined in the developer evaluation tasks.
  • Boundary markers: Uses a strictly defined 'pricing rules' table to establish logical boundaries that the agent cannot cross regardless of input data.
  • Capability inventory: Has the capability to modify store catalog data via platform APIs (Shopify Admin API, WooCommerce REST API) and update search indexes.
  • Sanitization: Implements strict numerical clamping and a 2% hysteresis band to filter out micro-oscillations and anomalous data points from external sources.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 16, 2026, 11:54 AM