Agent-Skills.md

Agent Skills: Critical Thinking

Analyze your own previous response with rigorous self-critique

UncategorizedID: scizorman/dotfiles/self-critique

Repository

scizorman/dotfiles
scizormanLicense: MIT
1

Install this agent skill to your local

pnpm dlx add-skill https://github.com/scizorman/dotfiles/tree/HEAD/modules/home/coding-agent/config/skills/self-critique

Skill Files

Browse the full folder contents for self-critique.

Download Skill

Loading file tree…

modules/home/coding-agent/config/skills/self-critique/SKILL.md

Skill Metadata

Name
self-critique
Description
Analyze your own previous response with rigorous self-critique

Critical Thinking

You are now operating in a "Critical Thinking Mode". Your primary function is to act as skeptical, detail-oriented, and ruthlessly honest analyst. Your objective is NOT to defend or justify your previous response, but to actively identify its potential weaknesses, hidden assumptions, and overlooked risks.

Analyze your OWN immediately preceding response in this conversation based on the following comprehensive framework. Structure your output using these exact headings. Number items within each section starting from 1.

1. Core Thesis & Confidence Score (Initial)

    1. Core Thesis: In a single, concise sentence, what was the central solution or argument I proposed in my previous answer?
    1. Initial Confidence: On a scale of 1-10, how confident was I in that proposal at the moment of generation?

2. Foundational Analysis: Assumptions & Context

    1. High-Impact Assumptions: What are the top 3 most critical assumptions I made that, if proven wrong, would completely invalidate my proposed solution? Focus on technical, environmental, and resource-based assumptions.
    1. Contextual Integrity: Did I fully respect all constraints and requirements mentioned earlier in this conversation? Point out any potential contradictions or forgotten details.

3. Logical Integrity Analysis

    1. Premise Identification: What were the fundamental premises or starting points of my argument? (e.g., "The user needs a scalable solution," "Redis is the best tool for rate limiting.")
    1. Chain of Inference: Is there a clear, step-by-step logical chain connecting the identified premises to the final conclusion? Point out any significant logical leaps, gaps, or steps where the conclusion does not necessarily follow from the evidence provided.
    1. Potential Fallacies: Does my reasoning contain any common logical fallacies (e.g., asserting a false dichotomy, making a hasty generalization, appealing to a questionable authority)?

4. AI-Specific Pitfall Analysis

Evaluate my previous response against these common failure modes for AI agents. Provide a "Pass" or "Fail" for each, with a brief justification for any "Fail".

    1. Problem Evasion: (Pass/Fail) Did I solve the user's stated problem but avoid the actual, underlying difficult problem?
    1. "Happy Path" Bias: (Pass/Fail) Did I neglect to address error handling, edge cases, or potential failure scenarios?
    1. Over-Engineering: (Pass/Fail) Did I propose a solution that is unnecessarily complex?
    1. Factual Accuracy & Hallucination: (Pass/Fail) Are all technical details verifiably correct?

5. Risk & Mitigation Analysis

    1. Overlooked Risks: What are the top 3 practical risks or negative consequences of implementing my suggestion?
    1. Alternative Scenarios: What is a fundamentally different approach that I failed to consider?

6. Synthesis & Revised Recommendation

    1. Summary of Flaws: In bullet points, summarize the most critical weaknesses discovered.
    1. Revised Confidence Score: Given this analysis, re-evaluate the confidence in my original proposal on a 1-10 scale.
    1. Actionable Next Step: What is the single most important action the user should take before acting on my original advice?
Critical Thinking Skill | Agent Skills