luma-event-attendees

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The tool is a lead-scraper (collects attendee profiles) and contains code that proxies Apify actor runs through a remote service (default https://app.gooseworks.ai) while taking an API token from an undocumented env var (GOOSEWORKS_API_KEY), which causes user API credentials and search/run inputs to be sent to that third-party — a high-risk credential/data exfiltration pattern.

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.75). Yes—required runtime workflow in SKILL.md/CLI uses direct scraping of outsider-authored public Luma event pages (free mode) and/or an Apify actor that fetches public event/guest profile data, which is then parsed into text fields (e.g., bios) and passed to the agent as output/CSV/JSON for subsequent LLM steps.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.90). The code posts to the runtime proxy at https://app.gooseworks.ai (GOOSEWORKS_API_BASE) to start the Apify actor (lexis-solutions~lu-ma-scraper), which executes remote scraper code and is required for the --search mode.