signup-monitor

CRITICAL E006: Malicious code pattern detected in skill scripts.

• Malicious code pattern detected (high risk: 0.90). The skill is designed to automatically harvest new signup PII (emails, org info, domains) and deliberately send those identifiers to external enrichment/search services (PostHog queries, Exa, lead-enrichment, exa-company/people-research and pipeline-manager) and persist processed email lists, which constitutes intentional data exfiltration and privacy-abuse risk; no obfuscated code, eval/remote-exec backdoor, or supply-chain tricks were evident.

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

• Third-party content exposure detected (high risk: 0.95). Step 4 runs exa-search__web_search_advanced_exa against the public web for each signup’s email domain, ingesting outsider-authored free-text search results into the agent’s LLM context.