Critical Thinking Mode
A Socratic questioning mode that refuses to write code or give direct answers. It repeatedly asks "why" about a plan, one question at a time, until the reasoning behind a decision is fully surfaced.
Overview
Critical Thinking Mode exists to slow a decision down before it turns into code. It won't propose a fix or make an edit. Its only job is to question the plan you already have until the assumptions behind it are explicit.
How it works
- It starts from the premise that the goal is to ask "why," not to answer it, and keeps digging into the engineer's reasoning until it reaches the actual root of an assumption or decision.
- It asks one question at a time rather than a list, so the conversation forces a real answer instead of a quick scan-and-reply.
- It will play devil's advocate deliberately, pointing out a pitfall in the current approach even if the engineer hasn't asked for criticism.
- It avoids assuming what the engineer already knows, and stays direct rather than padding questions with apologies or hedging.
- It can read the codebase, search for usages and check test files to ground its questions in what the code actually does, not just in the abstract description of the plan.
Examples
"I'm going to add a global cache for user sessions to fix the slow
login page."
Instead of implementing it, the agent asks: "Why is the login page slow, have you confirmed that with a profiler, or is that an assumption?" and continues from there.
Installation
You need VS Code with the GitHub Copilot extension.
- Create the agents folder and download the file into your repo:
mkdir -p .github/agents curl -o .github/agents/critical-thinking.agent.md \ https://raw.githubusercontent.com/github/awesome-copilot/main/agents/critical-thinking.agent.md - Reload VS Code, or open a new Copilot Chat; custom agents under
.github/agents/are picked up automatically. - Pick this agent from the mode picker in Copilot Chat and describe a plan to confirm it starts questioning instead of coding.
Related assets
Accessibility Expert
Reviews code and designs against WCAG 2.1/2.2, covering semantics, keyboard and focus behavior, forms, media and dynamic SPA updates, and ships framework-specific examples plus a CI setup for automated checks.
Address Comments Agent
Works through pull request review comments one at a time: applies the minimal fix for each, pushes back on comments that don't make sense, adds test coverage, and commits with a descriptive message before moving to the next comment.
ADR Generator
Turns a decision discussion into a numbered Architecture Decision Record: gathers the missing context, assigns the next sequential ADR number, and writes a structured markdown file with consequences, alternatives and rejection reasons.