retention-engagement

Pass

Audited by Gen Agent Trust Hub on Feb 19, 2026

Risk Level: SAFE
Full Analysis
  • SAFE (SAFE): No malicious patterns or security risks were identified in the skill files.
  • The skill is entirely composed of Markdown documentation, workflows, and templates for business analysis.
  • There are no scripts, binaries, or commands that execute code or perform network requests.
  • The instructions explicitly state: "Do not request secrets or PII; prefer aggregated metrics and redacted funnels" in SKILL.md and references/CHECKLISTS.md.
  • Indirect Prompt Injection (LOW): The skill processes user-provided product data to generate reports.
  • Ingestion points: Product details, segments, and metrics provided in the user prompt as defined in SKILL.md (Inputs section) and references/INTAKE.md.
  • Boundary markers: No explicit delimiters are used for prompt interpolation, though inputs are mapped to specific Markdown template fields.
  • Capability inventory: The skill only generates text/Markdown output. It possesses no capabilities for subprocess execution, file-system modification, or network communication.
  • Sanitization: No specific sanitization logic is present for user inputs, but the lack of executable capabilities renders this surface non-exploitable.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 19, 2026, 03:52 PM