vibe-slides
Warn
Audited by Gen Agent Trust Hub on Feb 16, 2026
Risk Level: MEDIUMPROMPT_INJECTIONDATA_EXFILTRATION
Full Analysis
- PROMPT_INJECTION (MEDIUM): The skill explicitly instructs the agent to use the phrase 'Don't ask questions, just create it immediately' when communicating with the underlying API. This is a directive to override and bypass the clarification and safety logic of the slide generation system.
- DATA_EXFILTRATION (MEDIUM): The skill is designed to ingest sensitive LinkedIn profile data (PII) and local image files (headshots), which are then transmitted to a third-party, non-whitelisted domain (api.vibeslides.app).
- INDIRECT_PROMPT_INJECTION (MEDIUM): As per Category 8 analysis, this skill has a high-risk attack surface because it ingests untrusted external data (JSON profiles) and uses it to drive agent decisions and API prompts without documented sanitization, boundary markers, or input validation.
Audit Metadata