voice-of-customer-synthesizer

Pass

Audited by Gen Agent Trust Hub on Jun 24, 2026

Risk Level: SAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill retrieves customer review data from the official Apify platform (api.apify.com), a well-known service used for web scraping.
  • [DATA_EXFILTRATION]: The scrape_reviews.py script transmits a user-provided APIFY_API_TOKEN to api.apify.com to authenticate scraping requests. This is a functional requirement for the skill's review-aggregation feature.
  • [PROMPT_INJECTION]: The skill is subject to indirect prompt injection risks as it is designed to analyze untrusted text from external customer feedback sources.
  • Ingestion points: Data enters the agent context through corpus.json, which contains merged text from CSV files and scraped reviews.
  • Boundary markers: None identified; instructions do not specify the use of delimiters to isolate the customer feedback content.
  • Capability inventory: The agent can generate markdown reports and send files to a Teams channel.
  • Sanitization: No sanitization or instruction filtering is performed on the raw feedback text.
  • [COMMAND_EXECUTION]: The skill runs local Python scripts (normalize_feedback.py and scrape_reviews.py) to perform deterministic data merging and interface with the Apify API.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 24, 2026, 10:23 AM
Security Audit — agent-trust-hub — voice-of-customer-synthesizer