Security audit
Security checks for vulnerabilities and agentic risk
This is a documentation-only workflow helper with no executable code, persistence, credential handling, or hidden data movement, though its activation wording is overly broad.
Before installing, be aware this skill may activate on broad self-improvement or logs-related language. It appears safe as a workflow helper, but users who want precise routing should narrow the trigger keywords or disable implicit invocation.
The keyword list includes extremely broad terms such as "self", "improving", and "own", which can match ordinary user requests unrelated to this skill’s intended scope. That creates an overbroad activation surface where the skill may be invoked unexpectedly, increasing the chance of prompt/context hijacking, incorrect workflow execution, or interference with more appropriate skills.
The manifest description says to use the skill when a user asks for broad categories like work-productivity, self, improving, logs, or any practical workflow or analysis support, which is too ambiguous to safely constrain invocation. This can cause the orchestrator or agent to select the skill for many unrelated requests, leading to misapplication of instructions and increased attack surface through unintended skill inclusion.
The trigger list includes extremely broad terms like "self," "improving," "logs," and "own," which are common in ordinary user prompts and can cause the skill to activate far outside its intended scope. Over-broad activation increases the chance of unintended routing, prompt-context pollution, and accidental application of this workflow in unrelated tasks, which is especially risky for a skill positioned to influence reliability and self-improving workflows.
The trigger keyword list includes extremely broad terms such as "self", "own", "improvement", and "bug fix", which are common in many unrelated conversations. This can cause the skill to activate unintentionally and steer the agent into an irrelevant workflow, increasing the risk of prompt-routing mistakes and unsafe over-application of the skill.
The trigger sentences and keywords are broad enough to match ordinary user language such as 'help me', 'I need a practical workflow', 'self', 'improving', or 'logs'. This can cause the skill to activate outside its intended scope, leading to inappropriate routing, accidental execution of the wrong workflow, and increased exposure to prompt-surface abuse if adversarial users intentionally craft ambiguous requests.
The trigger phrases are broad, generic, and overlap with ordinary user requests such as asking for a workflow, bug-fix help, or improvements. This increases the chance the skill is invoked unintentionally in contexts the user did not explicitly intend, which can route tasks through the wrong instructions and create unsafe or confusing behavior in a self-improving workflow setting.
The trigger phrases are written in broad natural language and can match many unrelated help requests because they lack clear boundaries on domain, task type, and expected artifact. In an agent ecosystem, ambiguous invocation logic is dangerous because it can cause accidental routing to this skill, leading to unintended handling of user data, misleading outputs, or disruption of safer specialized workflows.
The example trigger phrases are largely templates around generic requests like "help me" or "I need a practical workflow," without meaningful scope restrictions. Generic examples train matching systems and maintainers toward permissive activation behavior, reinforcing accidental invocation and making it easier for unrelated prompts to satisfy the trigger pattern.
The description says to use the skill when users ask about broad topics like work-productivity, self, improving, or logs, without clearly constraining the scenario. Because these terms are vague and overlap with many benign requests, the agent may invoke this skill in contexts where it does not fit, leading to misrouting, confusion, or inappropriate workflow generation.
The default prompt trigger phrase is extremely broad and composed of common terms like 'work-productivity,' 'self-improving-agent,' 'logs,' 'workflow,' and 'analysis,' which are likely to appear in ordinary user requests. This can cause unintended invocation of the skill, exposing users to unexpected prompt injection surface or causing the agent to act in a workflow it did not explicitly choose.
Enabling implicit invocation without clear activation constraints allows the skill to be auto-selected during unrelated or only loosely related conversations. In a self-improving workflow skill, this is more dangerous because the skill is positioned to influence debugging, safety, reliability, and adjacent implementations, increasing the chance of untrusted workflow changes or overreach from accidental activation.
