Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only helper for unit-test coverage work, with broad activation wording but no hidden execution, persistence, credential use, or destructive behavior.
Install this if you want a repeatable unit-test and coverage workflow. Be aware it may activate on broad testing or quality phrasing, so explicitly invoke it for unit-test coverage tasks and review any suggested code changes before applying them.
The trigger sentences are generic and permissive, making it easy for the skill to activate outside narrowly intended unit-testing contexts. Over-broad activation can cause misrouting, unexpected invocation, or prompt-surface expansion where the skill influences unrelated requests, which is a real security and reliability concern for agentic systems.
The trigger phrases are very broad and include generic terms like 'unit tests', 'test coverage', and 'testing', which can cause the skill to activate in situations where it was not specifically intended. Over-broad activation increases the chance of prompt hijacking, unintended invocation, or interference with other skills/workflows, especially in automated routing systems that rely on keyword or phrase matching.
The skill description and activation guidance are broad enough to match many routine software-help requests, which can cause the agent to invoke this skill outside its intended scope. Overbroad routing increases the chance of inappropriate tool or workflow selection, leading to confused-deputy behavior, irrelevant code changes, or unsafe assumptions in unrelated development tasks.
Generic keywords like 'software-and-data', 'testing', 'regression', and 'quality' are too common across developer workflows and can cause accidental activation for unrelated requests. In an agent system, this weakens skill isolation and can let broad user phrasing steer execution into a testing-oriented workflow that is not appropriate for the task.
The skill description defines activation conditions too broadly, covering wide software-and-data and testing-related requests without clear limits. This can lead to over-selection of the skill, reducing routing precision and causing users to receive generic unit-test guidance when a different, more appropriate skill should handle the task.
The trigger keywords are broad and weakly scoped, including generic terms like testing, regression, and quality. In an agent routing system, this can cause the skill to activate for loosely related requests, increasing the chance of inappropriate delegation, irrelevant guidance, or accidental exposure of project context to an unnecessary skill.
The default prompt contains broad trigger terms such as software-and-data, unit tests, test coverage, testing, regression, workflow, artifact, checklist, analysis, and implementation support. Combined with allow_implicit_invocation: true, this creates a realistic risk that ordinary user requests will unintentionally activate the skill, expanding its influence beyond clear user intent and potentially causing prompt-routing confusion or overbroad tool use.
The invocation guidance does not clearly define when the skill should and should not trigger, leaving boundaries ambiguous. In an agent environment, vague trigger scope can cause accidental or adversarial invocation, reducing routing integrity and allowing the skill to intercept common development requests it was not designed to handle.
The trigger sentence is broad enough to match ordinary user phrasing about needing help or a workflow, which can cause the skill to activate outside its intended scope. Overbroad activation can route unrelated requests into this skill, producing misleading test-coverage guidance and increasing the chance of prompt-selection abuse or unintended tool behavior in a larger agent system.
The file list specifies English and Chinese instruction variants, but the README does not state that users may choose their preferred language or locale. This can be read as a language-policy issue because the documentation presents fixed language variants without explicit opt-in guidance.
The example trigger sentences use everyday phrasing that effectively teaches the router to activate on vague requests without validating scope or constraints. This is less severe than executable abuse, but it still promotes unnecessary or incorrect invocation, which can degrade reliability and increase the risk of unsuitable outputs in adjacent software tasks.
