Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only testing helper with overly broad activation wording but no hidden execution, persistence, credential use, or unsafe behavior.
Installers should be aware this skill may activate on broad software quality or testing language. It is safest when invoked explicitly for adding unit tests, improving coverage, or creating regression-test workflows.
The skill description and activation guidance are broad enough to match many ordinary software-assistance requests, which can cause the skill to activate outside its intended scope. Over-broad routing increases the chance of unintended instruction injection, context hijacking, or suppression of a more appropriate specialized skill, especially in multi-skill agent environments.
The trigger sentence is so broad and awkwardly phrased that it can match ordinary user requests or loosely similar language, causing the skill to activate outside its intended scope. In an agent system, over-broad routing can expose users to irrelevant automated behavior, suppress better-matched skills, and create prompt-injection-style control issues through unintended invocation paths.
The trigger sentence is so broad and unnatural that it can activate on generic help-seeking phrasing rather than a clearly scoped request for this specific skill. In an agent environment, overbroad triggers can cause inappropriate routing, invoke the wrong capability, and expose repository or code-analysis context when the user did not explicitly ask for this skill.
The invocation pattern 'I need a practical workflow for ...' is a generic request form that can match many unrelated user intents, making this skill eligible when a different workflow-oriented skill should respond. This increases prompt-routing ambiguity and may lead to unintended execution paths or disclosure of context to a skill the user did not mean to invoke.
The trigger phrases are broad, natural-language prompts that can match ordinary user requests about testing and workflows, causing the skill to activate outside narrow intended scenarios. In an agent environment, over-broad invocation can misroute requests, override more appropriate skills, or cause unintended handling of code and repository context at scale.
Generic keywords such as "testing," "regression," and "quality" are common across many unrelated software conversations, so they are poor discriminators for safe skill selection. If used for routing, they can cause accidental invocation and expose the agent to irrelevant instructions or broaden the effective authority of this skill beyond unit-test coverage help.
The example trigger beginning with "Help me" is ambiguous because that phrase appears in countless harmless user requests and does not meaningfully constrain the domain. While not directly malicious, such examples encourage loose matching behavior and can contribute to over-triggering in systems that learn or mirror activation patterns from examples.
The description defines activation conditions using a very wide set of categories, including broad domains like 'software-and-data' and general testing-related requests, without clear exclusion criteria. This increases the chance of unintended invocation, which can lead to prompt-routing confusion, irrelevant outputs, and reduced trust in the agent's behavior.
The trigger keywords are broad, generic terms such as 'testing' and 'quality', which can cause the skill to activate in situations far outside its intended scope. This is dangerous because misrouting can inject irrelevant testing-oriented guidance into unrelated tasks, reducing reliability and potentially interfering with higher-priority or more appropriate skills.
The invocation text includes the phrase "help me," which is a common everyday expression and not specific enough to reliably constrain when this skill should activate. The manifest does not provide narrower trigger conditions, exclusions, or negative examples to reduce accidental invocation.
The evidence list includes Chinese-language sources and labels, but the document does not explain any locale-specific scope or offer a user language/locale choice. This can create a mild language-policy concern if the skill implicitly relies on or surfaces non-user-selected locale content.
No suspicious patterns detected.
