user-onboarding

Pass

Audited by Gen Agent Trust Hub on Mar 1, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection by capturing unvalidated user input and persisting it into structured Markdown files used by other system components.
  • Ingestion points: User responses gathered during Phase 1 (Identity), Phase 2 (Schedule), and Phase 3 (Motivation/Goals) are used to populate profile files.
  • Boundary markers: No specific delimiters or instructions are provided to the agent to treat the user input as untrusted data when writing to files.
  • Capability inventory: The skill explicitly defines file-write operations to several files under the ~/.openanalyst/profile/ directory (e.g., profile.md, availability.md, preferences.md, motivation-triggers.md).
  • Sanitization: There is no evidence of input validation, escaping, or sanitization logic to prevent a user from injecting Markdown formatting or instructions that could influence downstream tools like the 'Schedule Replanner' or 'Motivation Generator'.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 1, 2026, 12:37 AM