quote-post
Pass
Audited by Gen Agent Trust Hub on Aug 3, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses the directive 'CRITICAL: Auto-start on load' and 'Do not summarise' in SKILL.md to override standard agent behavior and force an immediate start of the workflow.
- [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by processing untrusted user input and incorporating it into a prompt template for external use. 1. Ingestion points: User-supplied captions are collected in Step 1. 2. Boundary markers: The user content is wrapped in double quotes in the Step 4 prompt template. 3. Capability inventory: The skill is restricted to text generation and lacks dangerous capabilities like file writing or network access. 4. Sanitization: No sanitization or safety filtering is applied to the user-provided caption before it is interpolated into the final prompt.
Audit Metadata