client-onboarding

Pass

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: Indirect Prompt Injection Surface. The skill is designed to ingest data from external web sources such as search results, LinkedIn profiles, and ad scrapers.
  • Ingestion points: Web fetch and scraping tools utilized in Phase 1 (SKILL.md).
  • Boundary markers: The instructions for synthesizing information in Phase 2 do not specify the use of delimiters for untrusted data or include directions to ignore embedded instructions.
  • Capability inventory: The skill writes research findings to the local filesystem and generates structured YAML blocks in HTML comments to guide downstream automation tools.
  • Sanitization: There is no documentation of explicit sanitization or validation of the content retrieved from external sources.
  • Assessment: Although an indirect prompt injection surface exists, it is consistent with the skill's intended use-case as an intelligence-gathering tool.
  • [SAFE]: No hardcoded credentials, unauthorized network exfiltration, or malicious persistence mechanisms were detected.
  • [SAFE]: The installation base command and the orchestration of other internal skills follow standard vendor patterns for the gooseworks-ai ecosystem.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 20, 2026, 06:25 PM
Security Audit — agent-trust-hub — client-onboarding