market-seo
Pass
Audited by Gen Agent Trust Hub on Sep 21, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill instructs the agent to execute a local Python script (
scripts/analyze_page.py) using a user-supplied URL as a command-line argument to collect webpage data. - [INDIRECT_PROMPT_INJECTION]: The skill processes data extracted from external, untrusted websites, which creates a vulnerability where malicious instructions could be embedded in the site's content or metadata.
- Ingestion points: The agent ingests JSON formatted data containing titles, meta descriptions, headings, and other webpage content in
SKILL.md(Core Workflow) andreferences/seo-operating-reference.md(Automated Analysis Inputs). - Boundary markers: No specific boundary markers or "ignore instructions" warnings are defined to separate the ingested external content from the skill's operational logic.
- Capability inventory: The agent uses the ingested data to generate an
SEO-AUDIT.mdreport and provide content strategy recommendations. It does not appear to use the ingested data for further command execution or system-level changes. - Sanitization: There is no mention of sanitizing or validating the content extracted from the target URLs before the AI processes it for the audit.
Audit Metadata