Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only image workflow helper with some overly broad trigger metadata but no evidence of data access, code execution, persistence, or destructive behavior.
Installers should be aware that the skill may activate more often than intended in environments that honor the broad Chinese trigger metadata or implicit invocation. Its actual behavior is limited to producing image-workflow guidance, prompts, QA checks, and retry plans.
The trigger keyword list is excessively broad, including generic terms like "nano", "banana", "pro", "generate", "edit", and "images" that commonly appear in ordinary user requests. This can cause unintended activation of the skill, leading to prompt-routing hijacks, inappropriate context injection, and unreliable agent behavior when unrelated tasks are misclassified as matching this skill.
The trigger sentence begins with a very broad everyday phrase ('Help me ...'), which can cause the skill to activate during ordinary conversation even when the user did not intend to invoke this specific skill. In an agent environment, unintended invocation can misroute tasks, override more appropriate skills, or cause workflow execution on irrelevant inputs, reducing reliability and potentially exposing downstream systems to unnecessary processing.
The invocation description says to use the skill when users mention broad terms or need general practical support around the demand, but it does not clearly bound what requests are in-scope. This ambiguity increases the chance that orchestration logic will activate the skill for loosely related prompts, causing overreach, incorrect tool selection, and accidental insertion of unrelated workflow guidance into normal conversations.
The example trigger phrases are written as broad natural-language requests that resemble ordinary user speech and repeat generalized demand statements rather than precise invocation patterns. This trains or encourages matching systems to fire on common phrasing, which can amplify false positives and make the skill easier to trigger accidentally across unrelated requests.
The skill enables implicit invocation without any visible trigger constraints, scoping rules, or confirmation requirements. That increases the chance the agent will auto-select this skill in loosely related contexts, causing unintended prompt-generation behavior, possible context leakage into the skill, and expanded attack surface for prompt-injection or workflow manipulation.
This trigger remains overly generic ('I need a practical workflow ...') and lacks specific conditions that distinguish intentional invocation from normal user requests. Because the skill is a general productivity helper, broad matching makes accidental activation more likely and increases the chance that unrelated user prompts are captured by this skill instead of a safer or more suitable one.
The entire skill description is written only in Simplified Chinese, including the title and usage guidance, with no indication that users may choose another language or locale. This can violate language-choice policy when a skill implicitly constrains interaction to one language without explicit opt-in or justification.
