Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only helper for unit test coverage work, with some broad activation wording but no hidden execution, persistence, credential use, or destructive behavior.
Before installing, be aware that this skill may trigger on broad testing or quality requests because implicit invocation is enabled. It appears safe for normal unit-test and coverage assistance, but users may prefer narrower activation wording if they want tighter control over when it is used.
The activation guidance lists keywords and trigger sentences but does not define clear boundaries for when the skill should not be used. In an agent setting, missing scope limits can lead to inappropriate auto-selection, causing the model to apply testing-oriented instructions in contexts where they are irrelevant, lower quality, or conflict with user intent.
The trigger sentence is broad enough to match many ordinary requests about testing or practical workflows, which can cause the skill to activate outside its intended niche. Over-broad activation increases the chance of unintended instruction injection into unrelated conversations and can displace safer or more appropriate skills.
The trigger phrases are broad natural-language requests such as 'Help me' and 'I need a practical workflow' combined with a long generic requirement description, which can cause the skill to activate outside clearly intended contexts. In an agent system, ambiguous activation boundaries can route unrelated user requests into this skill, leading to incorrect handling, prompt/skill hijacking opportunities, or unintended access to project context during testing-related workflows.
The skill description and invocation guidance use broad phrases like 'software-and-data', 'testing', and 'regression' that can match many ordinary engineering requests outside the narrow purpose of improving unit-test coverage. In an agent routing system, this increases the chance of unintended activation, causing the agent to apply the wrong workflow or disclose irrelevant internal artifacts in contexts where this skill was not explicitly requested.
The keyword list contains generic terms such as 'testing', 'regression', and 'quality' that are too ambiguous to safely distinguish this skill from many unrelated developer-assistance tasks. This creates a prompt-routing vulnerability where normal conversation may invoke the skill unexpectedly, leading to scope confusion and potentially unsafe autonomous behavior if the skill starts making assumptions about code or verification steps.
The example trigger sentences are phrased so generically that they overlap with ordinary help requests, which trains or encourages the router to treat common conversational wording as a valid activation signal. Because this is a testing-oriented skill with broad implementation guidance, accidental triggering is more likely in benign software discussions, increasing misrouting risk and reducing user control over tool selection.
The top-level description defines the skill scope very broadly across multiple common software topics and artifact types without clear exclusion criteria. In an agent routing system, this can cause the skill to be selected for loosely related requests, increasing misrouting risk and potentially displacing safer or more relevant skills with narrower responsibilities.
The trigger list includes very broad generic keywords such as "testing" and "quality", which can cause the skill to activate in unrelated contexts. Over-broad activation increases the chance that the agent applies this workflow when the user did not request it, which can lead to irrelevant guidance, context hijacking, or accidental interference with more appropriate skills.
The skill enables implicit invocation while advertising a very broad testing-related capability through a generic default prompt. This can cause the agent to trigger the skill in loosely related conversations about software, testing, or regression work, creating unintended context routing and increasing the chance that repository code, internal implementation details, or user requests are sent to the skill without clear user intent.
The trigger sentence uses a very broad everyday phrase ('Help me ...'), which can cause the skill to activate in contexts far beyond unit testing or coverage work. Overbroad invocation patterns increase the risk of unintended routing, causing the agent to apply this skill to unrelated requests and potentially generate misleading code, testing guidance, or workflow actions under the wrong context.
This file is a Chinese-language user-facing README, but there is no statement that language selection is optional or that the locale is intentionally limited for a region-specific use case. Under the policy, forcing a specific language without user opt-in can be a natural-language policy concern.
The evidence list includes Chinese-language source titles alongside English content, but the file does not state whether the skill operates in multiple languages or how language selection is handled. This can create ambiguity around language or locale expectations for users.
