Security audit
Security checks for vulnerabilities and agentic risk
This is a simple Markdown-only learning workflow skill with broad activation phrases but no executable behavior or sensitive access.
Install only if you want a lightweight Speak-related reference skill. Expect possible accidental activation from the common word "speak"; narrowing the trigger would improve usability, but no dangerous behavior was found in the artifact.
The trigger field includes very broad phrases such as "speak" and "Speak", which are common words likely to appear in ordinary conversation. This can cause unintended activation of the skill in unrelated contexts, increasing the chance of confusing behavior, incorrect routing, or accidental invocation of low-trust workflow content.
The activation scenarios are ambiguous because they allow invocation whenever a user says generic terms like "speak" or requests vaguely related help. In a conversational system this broad matching can be exploited or accidentally triggered by benign user input, causing unintended skill selection and reducing trust boundaries between unrelated tasks.
