Security audit
Security checks for vulnerabilities and agentic risk
This text-only thinking skill has no code or persistence, but its broad shortcut-oriented guidance includes examples that can encourage bypassing ordinary rules or procedures.
Review this skill before installing if you do not want the assistant to suggest rule-bending shortcuts. Use it only for low-stakes, lawful, authorized experimentation, and avoid applying it to forms, queues, workplace rules, finance, medical, legal, safety, security, or other integrity-sensitive situations unless the boundaries are made explicit.
The examples 'ask who you could ask to skip it' and 'ask what happens if you leave fields blank' are concrete suggestions for circumventing procedures and submitting incomplete information. Because they are framed as everyday applications of the skill, they materially increase the chance that users interpret the skill as endorsing procedural evasion or dishonest conduct.
Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
Ask: If I asked an expert how to solve this, what would they say?
Write down the standard method step by step. Do not judge it yet. Simply document it.
**Example - Learning Photography:**
- Buy a DSLR camera
The README frames 'xiexiu thinking' as a broadly applicable meta-skill for finding shortcuts in any domain and repeatedly valorizes bypassing orthodox methods. Although later sections include some caveats, the opening positioning lacks immediate guardrails and can normalize cutting corners before readers encounter limits, increasing the chance of misuse in compliance-, safety-, or ethics-sensitive contexts.
The daily mindset examples encourage testing whether one can skip queues by asking the right person and leaving form fields blank to see what happens. Those suggestions can be interpreted as probing or bypassing procedural controls, which may facilitate unfair access, fraud, or evasion of required information, especially because no permission/compliance boundary is stated alongside the examples.
The description says to use the skill when users 'need help developing lateral thinking, first principles reasoning, or creative approaches to problems,' which is a very broad condition that overlaps with many ordinary assistance requests. It does not provide clear boundaries or exclusion conditions for when the skill should or should not activate beyond a few examples.
The core framing praises rejecting orthodox methods, forbidden techniques, and absurdly efficient paths without sufficiently strong guardrails. In a general-purpose assistant skill, this can steer users toward bypassing rules, constraints, or governance in domains where such shortcuts are unsafe or improper.
The documentation explicitly states that xiexiu is 'not a collection of tricks.' Later sections nevertheless present direct tactic prompts like skipping queues and omitting form fields, which contradict the stated intent to avoid prescribing specific hacks.
The skill says it teaches a general way of thinking, but the daily-life examples cross into actionable prompts for bypassing ordinary processes, such as skipping queues or omitting form fields. That framing can normalize procedural evasion and encourage users to apply the methodology in unethical or policy-violating contexts.
No suspicious patterns detected.