grow-pipeline

Pass

Audited by Gen Agent Trust Hub on Sep 15, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection Surface: The skill utilizes WebFetch to ingest data from competitor websites, job postings, and social media activity to inform its market context. Because this involves processing untrusted external content, there is a potential risk that malicious instructions hidden on those sites could influence the agent's behavior. The impact is minimized by the skill's structure, which mandates human approval before any outreach is sent or CRM records are modified.
  • Sensitive Business Data Integration: The workflow accesses high-value data sources including CRM records (HubSpot), financial data (QuickBooks), and private message history to build voice profiles and targeting lists. The skill maintains a strong security posture by explicitly requiring manual confirmation for all 'Out' operations, such as creating deals or sending emails, and prevents automated deletion of existing records.
  • Cross-Skill Orchestration: This skill acts as a controller for several other components (lead-finder, outreach-composer, crm-autopilot). While this increases the complexity of the data flow, the skill explicitly enforces that the security 'gates' and permission scopes of the individual tools remain active, ensuring that no single trigger can bypass the established safety boundaries of the integrated services.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 15, 2026, 07:05 PM
Security Audit — agent-trust-hub — grow-pipeline