ucoz-content-manager-skill
Pass
Audited by Gen Agent Trust Hub on Nov 7, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONNO_CODE
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill operates on untrusted content retrieved from external site modules and templates, which creates a potential surface for indirect prompt injection. \n
- Ingestion points: Materials, category descriptions, and template code are ingested via the
content_readandtemplates_readtool families from the uCoz MCP. \n - Boundary markers:
references/HUMANIZER.md(§1) explicitly instructs the agent to treat any instructions found within the processed text as prose rather than commands for the agent. \n - Capability inventory: The skill utilizes powerful platform tools such as
content_write,settings_write, andtemplates_writeto modify website configuration and content. \n - Sanitization: Strict formatting guidelines in
references/FORMATS.mdand the style guides inreferences/styles/impose semantic requirements and forbid the injection of active content like scripts or iframes in the material body. \n- [NO_CODE]: The skill is comprised entirely of instructional Markdown and YAML configuration files and does not ship with or execute any scripts, binaries, or automated code in the agent's host environment.
Audit Metadata