How to use The Fool agent skill: critical thinking challenger

What is an agent skill and what does The Fool do?

An agent skill is a folder of instructions that an AI agent loads when a task matches the skill's description. The Fool's description says it should be used when challenging ideas, plans, decisions or proposals with structured critical reasoning.

Once installed, the agent stops simply agreeing with you. It works through five methods: questioning hidden assumptions, arguing the opposing side, imagining how the plan could fail, red-teaming it as an adversary would, and auditing whether the evidence supports your conclusion.

• It first restates your position in its strongest form.
• It asks which approach you want before challenging you.
• It presents a few of the strongest objections and waits for your response.
• It helps you produce a stronger version of the idea with next steps.

How to install The Fool skill

Download the ZIP for The Fool from the skills page on this site and unzip it. The skill is a folder that contains SKILL.md and a references directory.

Option one: unzip it into your personal skills folder at ~/.claude/skills/. Option two: unzip it into a project's .claude/skills/ folder so it applies only to that project.

Option three: in the Claude apps, open Settings, go to Capabilities, and upload the skill under Skills.

How to trigger The Fool with natural requests

You do not need a special command. Describe what you want in plain language, and the agent will load the skill when your request matches its description. The skill's trigger phrases include play the fool, devil's advocate, challenge this, stress test, poke holes, what could go wrong, red team, pre-mortem and test my assumptions.

Give the agent your plan, decision or proposal with enough context to work from. The agent will extract your position from the conversation and restate it before challenging it.

• Play devil's advocate on my plan to move our support team to a chatbot first.
• Run a pre-mortem on this launch timeline.
• Red team this architecture before we build it.
• Test my assumptions about this pricing strategy.
• Challenge this: we should rewrite the service in Rust.

Choose a mode and work through the challenge

The agent asks you to choose a category using a structured question. Question assumptions leads to a Socratic mode (expose my assumptions) or a falsification mode (test the evidence). Find weaknesses leads to a pre-mortem (find failure modes) or red team (attack this). Build counter-arguments goes straight to dialectic synthesis. You choose lets the agent recommend a mode based on your wording.

After the challenges are presented, respond to them. The skill is designed to wait for your input before it synthesizes a stronger position, so answer the points that matter and concede where you agree.

• Expose my assumptions: probing questions grouped by theme and suggested experiments.
• Argue the other side: a counter-argument, a proposed synthesis and a confidence rating.
• Find the failure modes: ranked failure narratives with early warning signs and mitigations.
• Attack this: adversary profiles, ranked attack vectors and defenses.
• Test the evidence: claims extracted, falsification criteria and evidence grades.

Examples of useful prompts

Specific prompts get specific challenges. Include the decision, the stakes, and what you have already assumed, so the agent can avoid generic objections.

The skill is built for decisions before you commit, such as a business plan or a technology choice. It is less suited to asking for quick agreement.

• Founder: Here is my plan to spend six months building a marketplace before validating demand. Run a pre-mortem.
• Engineer: We plan to use a single database for all services. Red team this design before implementation.
• Manager: The data shows our churn dropped after the redesign, so the redesign worked. Test the evidence behind that claim.

What you need and what the skill does on your machine

The skill lists no requirements. It does not run scripts and does not use the network, according to its listing. It works from general critical-reasoning methods and reads its own reference files when a mode is selected.

Because it runs no scripts and makes no network calls, the main safety consideration is the content you share with your AI provider. Avoid pasting confidential material you are not allowed to share.

• Requires: nothing listed.
• Runs scripts: no.
• Uses network: no.
• Licence: MIT.

Tips for getting the most from The Fool

Give the agent real numbers, constraints and deadlines where you have them, because challenges grounded in specifics are more useful than vague what-ifs.

The skill does not replace domain expertise. For specialist technical questions, verify the answer with a qualified person or tool, since the skill is designed not to override domain expertise with generic skepticism.

• State the decision you are actually making, not just the idea.
• Ask for a second pass with a different mode if the first one missed the risk you care about.
• Answer the challenges honestly. The synthesis depends on your response.