ui-loop
Warn
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: MEDIUMPROMPT_INJECTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill uses a custom 'Prompt DSL' containing the directive
::OBJECTIVE{site_design|pri:OVERRIDE_ALL}. This uses an explicit override marker (OVERRIDE_ALL) intended to prioritize the skill's instructions over system-level constraints or prior user directions. - [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read and analyze untrusted external data, which could contain hidden malicious instructions.
- Ingestion points: The agent is directed to read the 'current page' and compare it against 'BENCHMARK: Awwwards | Webby Awards | FWA winning entries' (external websites).
- Boundary markers: There are no boundary markers or instructions to ignore embedded prompts within the processed UI content or the benchmark websites.
- Capability inventory: The agent has the capability to modify the current page (
[PATCH]) and loop execution based on the ingested content. - Sanitization: The skill lacks any instructions to sanitize or validate content fetched from external UI benchmark sites before it influences the agent's actions.
Audit Metadata