tech-blog-seo-rewriter
Fail
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: HIGHPROMPT_INJECTION
Full Analysis
- PROMPT_INJECTION (HIGH): The skill is highly vulnerable to Indirect Prompt Injection due to its data ingestion patterns combined with file-writing permissions.
- Ingestion points: In
FORMS.md, the skill defines workflows to ingest untrusted article content from local paths or URLs, and search performance data from CSV files and screenshots. - Boundary markers: No delimiters, markers, or explicit 'ignore embedded instructions' warnings are present in
SKILL.mdorFORMS.mdto prevent the agent from obeying commands contained within the content it is rewriting. - Capability inventory:
SKILL.mdauthorizes the agent to write and save optimized files to thetech-blog/output/directory, providing a side-effect vector for successful injection. - Sanitization: There are no instructions or mechanisms defined to sanitize, escape, or validate the external content before it is processed by the agent.
Recommendations
- AI detected serious security threats
Audit Metadata