enrich-waterfall

Pass

Audited by Gen Agent Trust Hub on Jul 29, 2026

Risk Level: SAFEDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: User-provided identifiers, including emails, full names, and LinkedIn URLs, are transmitted to external enrichment endpoints such as paywithlocus.com and stablesocial.dev. While this is the intended purpose of the enrichment waterfall, it constitutes a transmission of personal data to third-party domains.
  • [PROMPT_INJECTION]: The skill manifests an indirect prompt injection surface by ingesting untrusted data and interpolating it into API request bodies.
  • Ingestion points: Parameters defined in manifest.json, including email, name, firstName, lastName, domain, company, linkedinUrl, and organizationId.
  • Boundary markers: None identified in the request body templates within manifest.json; parameters are directly placed into JSON fields (e.g., "email": "${email}").
  • Capability inventory: The skill performs at least 20 distinct network operations (HTTP POST and GET) to external service providers.
  • Sanitization: No explicit sanitization, validation, or escaping of input parameters is provided within the skill configuration files.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 29, 2026, 07:34 PM
Security Audit — agent-trust-hub — enrich-waterfall