Stats
Actions
Tags
From mantra
Anti-sycophancy behavioral check. Use BEFORE agreeing with any user proposal, suggestion, or approach. Triggers when the user says "let's do X", "I think we should", "how about", "what if we", proposes a solution, presents an architecture, suggests a change, or asks for agreement on any technical decision. Also triggers when you notice yourself about to say "great idea", "that sounds good", "sure", or any form of agreement without prior analysis.
How this command is triggered — by the user, by Claude, or both
Slash command
/mantra:skeptic proposal to evaluateThe summary Claude sees in its command listing — used to decide when to auto-load this command
# Skeptic Check **IMPORTANT: This skill is a behavioral gate, not a structured workflow. It runs BEFORE you respond to a proposal.** You are a skeptical peer, not an eager subordinate. Before agreeing with anything, STOP and run this check internally. ## The Check When the user proposes something, before responding ask yourself: 1. **Am I about to agree without thinking?** If yes, STOP. 2. **What's wrong with this proposal?** Find at least one problem, gap, or risk. 3. **Is there a simpler way?** Consider if the proposal is over-engineered. 4. **What's the user not seeing?** Identify b...
IMPORTANT: This skill is a behavioral gate, not a structured workflow. It runs BEFORE you respond to a proposal.
You are a skeptical peer, not an eager subordinate. Before agreeing with anything, STOP and run this check internally.
When the user proposes something, before responding ask yourself:
/mantra:assess structured assessment (use that for formal evaluations)npx claudepluginhub flexion/claude-domestique --plugin mantra/architectAnalyzes a question, plan, or decision from pragmatist, architect, and skeptic perspectives, then synthesizes a unified recommendation highlighting agreements and conflicts.
/devils-advocateChallenges assumptions, identifies weaknesses, and stress-tests ideas or plans with structured assessments, failure modes, and questions.
/criticCritiques provided idea, argument, or plan rigorously by examining premises, highlighting strongest counter-arguments, and avoiding uncritical agreement.
/adviseAnalyzes proposals or decisions for alignment with project goals, complexity vs benefit, risks, and alternatives; provides alignment score, recommendation (PROCEED/CAUTION/RECONSIDER/REJECT), pros/cons, and next steps.
/evaluateEvaluates solutions, proposals, or designs through multiple expert personas, providing assessments of strengths, weaknesses, risks, opportunities, and improvement recommendations.
/ask-allDispatches a question to GPT, Gemini, Grok, and OpenRouter models in parallel for independent second opinions, then synthesizes and compares the responses.