Security audit
Security checks for vulnerabilities and agentic risk
This is a low-impact workflow-helper skill with overly broad auto-invocation triggers, but no hidden code, credential access, persistence, or destructive behavior.
Install only if you want a general ontology/typed-workflow planning helper. Be aware it may be invoked for broad terms like knowledge, graph, memory, or bug fix unless the platform lets you restrict implicit invocation.
The skill description and usage guidance are broad enough to match many ordinary requests involving workflows, analysis, implementation support, or generic productivity topics. In agent routing systems, overly broad triggers can cause incorrect skill selection, leading the agent to apply the wrong instructions or expose users to irrelevant or lower-quality behavior at scale.
The keyword list includes highly generic terms such as 'knowledge,' 'graph,' 'structured,' 'memory,' 'creating,' and 'bug fix' without any scope constraints. These terms are common across many unrelated domains, so an agent may invoke this skill for requests far outside its intended purpose, increasing the chance of misrouting, instruction interference, and unreliable outputs.
The trigger sentence is broad enough that ordinary user phrasing could activate this skill outside its intended scope, causing inappropriate routing or over-application of the workflow. In an agent ecosystem, loose activation criteria can become a security and reliability issue because the skill may process unrelated requests, override better-matched skills, or expose users to unsafe assumptions in contexts it was not designed for.
The trigger examples are broad, generic, and partially match common user phrasing such as asking for 'a practical workflow' or help with 'bug fixing' and 'hardening.' In an agent ecosystem, this increases the chance the skill is invoked unintentionally for loosely related requests, which can misroute tasks, cause unexpected behavior, and expand the skill's operational footprint beyond what the user intended.
The trigger phrases are extremely generic and can match a wide range of unrelated requests, causing the skill to activate outside its intended scope. In an agent ecosystem, overbroad activation can lead to unintended workflow execution, confusing results, or routing sensitive user tasks into a skill that was not explicitly requested.
The example trigger sentences are built around very generic lead-ins like 'Help me' and 'I need a practical workflow,' which trains or signals downstream routing logic to associate this skill with commonplace requests. Even if unintentional, this broadens activation beyond the stated domain and can degrade safety and correctness by overshadowing more appropriate specialized skills.
The trigger keywords are very broad and include common terms like "knowledge", "graph", "structured", and "bug fix", which can match many unrelated user requests. This can cause the skill to activate outside its intended scope, leading to prompt hijacking of normal conversations, reduced reliability, and accidental suppression of more appropriate skills or safeguards.
The example trigger phrases use vague, everyday wording such as "Help me" and "I need a practical workflow," which are broadly applicable to many normal requests. If used by an agent router or dispatcher, these examples can train or bias matching toward over-selection of this skill, causing unintended invocation and unsafe task routing.
The default prompt is extremely broad and maps to generic help-seeking language such as 'help me,' which can cause the skill to be invoked outside its intended scope. This creates prompt-routing confusion and raises the chance that unrelated user requests are implicitly steered into this skill, where its instructions could override or distort safer or more appropriate handling.
Enabling implicit invocation without tight activation constraints allows the platform to auto-select this skill based on vague similarity rather than explicit user intent. In combination with the broad prompt text, this increases the risk of unrequested activation, misrouting, and accidental application of the skill’s instructions to unrelated tasks.
A file presented as a Chinese README includes substantial English instructional text, and the trigger examples are also written in English. This can effectively impose a language preference without explicit user opt-in or a documented language-selection mechanism.
