Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only image workflow helper with no executable code, credential use, persistence, or hidden data movement.
Before installing, understand that this skill may be invoked for fairly broad AI image workflow requests. It appears safe to use, but users who maintain many overlapping image or workflow skills may want narrower trigger wording to avoid irrelevant responses.
The trigger sentences are broad, natural-language phrases that can match ordinary user requests unrelated to the intended skill boundary, increasing the chance of accidental invocation. In an agent environment, unintended activation can route user tasks into the wrong workflow, causing confusing behavior, prompt-scope expansion, and misuse of the skill outside its reviewed purpose.
The trigger example is broad enough to match many ordinary requests about AI image generation workflows, which can cause the skill to activate outside a narrowly intended scope. Over-broad routing increases the chance of inappropriate tool selection, prompt interception, or unreviewed influence over user tasks that only loosely relate to Nano Banana-specific workflow support.
The description uses very broad activation criteria such as 'needs practical workflow, code, checklist, documentation, or review support for this job,' which can cause the skill to trigger for loosely related requests. Over-broad routing is a real security and reliability issue because it can inappropriately insert this skill's instructions into unrelated contexts, increasing the chance of prompt interference, policy bypass attempts, or unsafe task handling outside the skill's intended scope.
The skill description is scoped broadly enough to match generic requests about image generation, editing, documentation, checklists, code, or review support, which increases the chance of accidental invocation outside the intended Nano Banana workflow context. Over-broad activation can route unrelated user tasks into this skill, causing inappropriate instructions, unnecessary data exposure to the skill context, or degraded safety/relevance guarantees.
The trigger section relies on a short list of broad keywords such as 'AI image generation', 'image editing', and 'visual QA' without any boundary conditions, making the skill likely to activate on many unrelated requests. In practice this can create routing confusion and unintended skill execution, especially because several terms are common across benign design, debugging, and review conversations.
The keyword list includes generic terms such as 'generate', 'edit', 'images', and 'bug fix', which are common across many unrelated requests and can over-expand the skill's trigger surface. This makes accidental or inappropriate invocation more likely, especially in multi-skill routing systems where broad keywords can overshadow more precise skills.
