Security audit
Security checks for vulnerabilities and agentic risk
This skill is a coherent diagram-making helper that edits local notes and images as disclosed, with only a minor risk of broad automatic invocation.
Install this if you want an agent to create and update diagrams inside your notes. Be aware that it may edit nearby note text and add image files when invoked, and the broad triggers mean you should watch for accidental use on ambiguous requests like general model comparisons.
The description lists triggers such as "draw a diagram" and especially "explain this visually," which are broad phrases likely to overlap with common user requests outside this skill's intended scope. The file does not provide negative examples or tighter activation constraints to distinguish when the skill should and should not activate.
The skill enables implicit invocation, allowing the agent to call it automatically based on broad user phrasing rather than explicit user consent. Because this skill can generate and visually verify rendered diagrams, automatic triggering may expose user content to an auxiliary tool unexpectedly and can cause the model to take side-effectful actions without a clear, narrow activation boundary.
The invocation scope is underspecified because phrases like 'compare these models' and 'explain this visually' can apply to a wide range of general tasks, not just diagram-making. In agent environments, this ambiguity can cause over-broad routing, where the skill is selected for requests involving sensitive, unrelated, or non-visual content, increasing the risk of inappropriate processing and user confusion.
The trigger phrases are broad enough to match many ordinary user requests unrelated to this specific skill, increasing the chance of unintended auto-invocation. Because this skill can create or modify diagrams and SVGs, accidental activation could expose unrelated user content to the skill workflow or cause the agent to take diagram-editing actions when not clearly requested.
No suspicious patterns detected.
