smb-onboard

Pass

Audited by Gen Agent Trust Hub on Jun 18, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill establishes an indirect prompt injection surface by collecting and persisting untrusted user input into a shared session memory block. This data is intended to be used by other skills to personalize their behavior, which could lead to a compromise if a user embeds malicious instructions within their business profile. \n
  • Ingestion points: User responses to onboarding questions regarding business sphere, size, and pain points in SKILL.md and reference/onboard-checklist.md. \n
  • Boundary markers: The information is stored using standard markdown headers (e.g., ## Контекст бизнеса) and bullet points, but lacks explicit delimiters or instructions to the agent to ignore embedded commands. \n
  • Capability inventory: The skill demonstrates the capability to write to 'Cowork session memory' and explicitly states that this context will be consumed by other skills. \n
  • Sanitization: The instructions include a manual approval gate where the agent must present the gathered information to the user for confirmation before it is saved to memory.
Audit Metadata
Risk Level
SAFE
Analyzed
Jun 18, 2026, 01:59 AM
Security Audit — agent-trust-hub — smb-onboard