The Agent Skills Directory

[PROMPT_INJECTION]: The skill uses imperative language and authoritative markers such as "CRITICAL OUTPUT RULE", "HARD OUTPUT RULE", and "CORE DIRECTIVE". It explicitly includes instructions such as "This rule overrides any model default," which is a common pattern in prompt injection used to force the agent to prioritize specific skill instructions over its standard operating parameters.
[PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection due to its processing of untrusted user input. Ingestion points: user-provided design briefs are interpolated into the skill's configuration logic (e.g., Section 1). Boundary markers: no explicit delimiters or "ignore instructions" warnings are present to isolate user data. Capability inventory: the skill's functionality is primarily restricted to image generation; no evidence of subprocess execution, file system access, or network operations was found. Sanitization: the skill lacks any mechanisms to validate or escape user input before it is processed by the model.

imagegen-frontend-web