ai-vibe-detector
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from user-provided URLs or pasted text to perform vibe analysis. It lacks structural boundary markers to prevent the agent from interpreting instructions embedded within the analyzed content, which is a vector for indirect prompt injection.
- Ingestion points: The Core Workflow in SKILL.md explicitly supports URL input and text extraction for analysis.
- Boundary markers: The instructions do not define any delimiters or ignore-instructions to isolate the analyzed text from the agent's control logic.
- Capability inventory: The skill triggers web scraping and browser automation tools to retrieve external content based on user input.
- Sanitization: No sanitization or validation of the fetched content is specified before it enters the agent's context.
Audit Metadata