Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only helper for unit-test coverage work, with no hidden code execution or persistence, though its implicit triggers are broader than ideal.
Installers should understand that this skill may activate on broad testing or quality-related requests because implicit invocation is enabled. It appears safe as a unit-test coverage helper, but users who want precise routing should narrow the trigger language before publishing broadly.
The invocation guidance defines keywords and trigger sentences without clear boundaries for when the skill should or should not activate. This increases the chance of accidental invocation from overlapping software requests, which can degrade routing reliability and expose downstream workflows to unintended execution paths.
The trigger sentence is phrased so generically that ordinary user requests about getting help or needing a workflow could unintentionally invoke this skill outside its intended scope. In an agent-routing system, broad triggers can cause mis-selection, leading the agent to apply this skill when a more appropriate or safer skill should handle the request.
The trigger phrases are broad and include generic wording like 'help me' and 'I need a practical workflow', which can cause the skill to activate in contexts far beyond the intended unit-testing scope. In an agent system, unintended invocation can route user requests to the wrong skill, leading to irrelevant actions, confusion, or unsafe chaining with other capabilities.
The skill description uses broad activation terms such as 'software-and-data', 'testing', and 'analysis or implementation support', which can match many unrelated requests. This creates an overbroad invocation surface where the wrong skill may activate, causing misrouting of user requests, irrelevant actions, or unintended access to project context during agent workflows.
The listed trigger keywords include vague terms like 'testing', 'regression', and especially 'quality', which are highly collision-prone across many software tasks. In an agentic environment, generic triggers increase the chance that this skill is selected for unrelated requests, leading to incorrect guidance or unnecessary handling of codebase context.
The example trigger sentences begin with generic phrases like 'Help me' and 'I need a practical workflow', which are common across many unrelated prompts. When paired with a broad requirement statement, they can train or encourage invocation patterns that are insufficiently specific, increasing accidental skill activation and prompt-routing errors.
The skill description and usage criteria are broad enough to activate on common software topics like testing, regression, or software-and-data without clear boundaries. That can cause over-triggering, where the skill is selected for loosely related requests and steers the agent into an unintended workflow, reducing routing accuracy and potentially suppressing more appropriate specialized skills.
The keyword list includes generic terms like 'testing' and 'quality' with no contextual constraints, which increases accidental invocation across many unrelated requests. In an agent environment, overly generic triggers can lead to prompt-routing hijack by matching broad user phrasing rather than actual intent, producing incorrect or lower-safety behavior.
The skill metadata uses very broad, everyday-language trigger terms such as software, unit tests, test coverage, testing, workflow, checklist, analysis, and implementation support. When paired with implicit invocation behavior, this can cause the agent to activate in loosely related conversations and inject behavior the user did not explicitly request, increasing the risk of prompt-scope confusion and unintended tool routing.
Enabling allow_implicit_invocation without strong trigger constraints allows the platform to auto-select this skill even when the user's request is ambiguous. In practice, this increases the chance of unintended skill execution, context hijacking within normal software conversations, and over-broad application of the skill beyond its intended unit-testing scope.
The trigger sentence is written in a very broad, natural-language form ('Help me ...') that can match ordinary user requests outside the intended narrow use case. This creates an overbroad activation surface, increasing the chance the skill is invoked unexpectedly and influences responses in contexts where it was not specifically requested.
The invocation description allows activation on broadly phrased requests for a 'practical workflow' without sufficiently constraining the subject matter, which can cause the skill to capture general software-assistance prompts. In an agent environment, overly permissive routing can misapply the skill, override better-matched behaviors, or expose users to unintended guidance paths.
This file is labeled as a Chinese README, but core user-facing content such as the requirement, usage description, and trigger phrases remains in English. That can effectively impose a language choice on users without opt-in or explanation, which is a locale-policy concern.
No suspicious patterns detected.
