sales-company-research

Pass

Audited by Gen Agent Trust Hub on Jun 16, 2026

Risk Level: SAFE
Full Analysis
  • Data Ingestion and Memory Management: The skill is designed to search through a variety of internal repositories—including wikis, CRMs, and messaging platforms—to build a repository of useful Sales resources. These findings are stored in persistent memory files for later use. This functionality introduces a potential surface for indirect prompt injection if external data contains conflicting instructions, though the skill includes logic to summarize and validate resources before saving.\n
  • Ingestion points: Wikis, CRM systems, shared drives, and internal communication channels mentioned in the workflow.\n
  • Boundary markers: The skill relies on natural language processing to evaluate sources rather than strict technical delimiters.\n
  • Capability inventory: Writing to local state and memory files such as plugin-memory.md and user-context.md.\n
  • Sanitization: Specific instructions to extract concise usage rules and redact sensitive details are included to clean data before persistence.\n- Cross-Skill Coordination: The workflow utilizes a modular approach by invoking and executing a preflight script from a related context skill. This allows for centralized management of user preferences and environment configuration across different Sales-related tasks.\n- Operational State Tracking: The skill maintains internal state by updating JSON and Markdown files. This ensures that the research process can track saved resources and pending questions across multiple sessions.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 16, 2026, 06:13 AM
Security Audit — agent-trust-hub — sales-company-research