conference-speaker-scraper

Pass

Audited by Gen Agent Trust Hub on Oct 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • [EXTERNAL_DOWNLOADS]: The skill connects to the Apify API (api.apify.com) to perform web scraping for JavaScript-heavy websites when the --mode apify flag is used.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external conference websites provided via the --url parameter, which could contain instructions designed to influence the agent.
  • Ingestion points: External HTML content is fetched in scripts/scrape_speakers.py from the user-provided URL.
  • Boundary markers: The skill output (JSON, CSV, or Summary) does not include specific delimiters or warnings instructing the agent to ignore embedded instructions within the scraped speaker bios or names.
  • Capability inventory: The skill itself performs network reads and prints data; however, the agent consuming this output may have broader system access or tool capabilities that could be targeted.
  • Sanitization: The extract_text function in the script uses regular expressions to remove HTML tags, but it does not sanitize or filter the actual text content for malicious instructional patterns.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 2, 2026, 07:11 PM