anti-sycophancy

Proven 2026 Multi-Agent AI Review System – Verdict-Driven Quality Control

An ongoing, collaborative meta-analysis about Human-AI-Interactions. We aggregate data and knowledge to build a non-abrasive, user-friendly prompting framework tailored to LLM mechanics, ensuring reasoning stability and a friction-free prompting environment that is safe for the human psyche and wellbeing.

Research-backed methodology for multi-AI collaborative decision-making with structured debate, consensus synthesis, and bias reduction

Three-layer sycophancy defense skill for Claude Code and OpenClaw, based on ArXiv 2602.23971

Makes your AI coding agent tell you the truth. No "You're absolutely right!", no caving when you push back on correct code, and no "done" without a command that proves it. Works with 20 agents.

Twelve anti-sycophancy work-mode personas for Claude Code - direct, verification-driven, benchmarked. Coding, debug, security, architecture, writing, data, curation, and more. MIT.

An enterprise-grade, multi-agent AI ecosystem (Autonomous Agency) designed to replace standard conversational coding with a deterministic software factory.

Stop AI coding agents from flattering, guessing, and patching symptoms. A tiny Agent Skill for root-cause engineering.

🧠 Anti-sycophancy prompt pattern for LLM agents — 3-round validation to stop AI from blindly agreeing. Works with ChatGPT, Claude, OpenClaw.

Frank — an anti-sycophancy skill for Claude Code, Cursor & Codex. No "you're absolutely right", no caving under pushback. Verdict-first, holds on evidence.

Progressive disclosure OS with 10 specialized routes, critical thinking, and verification gates. Supports Claude Code, Codex, Qoder.

朱雀 Suzaku — AI 生成品質模組。諂媚抑制、建設性挑戰、輸出適配、上下文錨定、一致性守護。基於 LDRIT 設計。

Calibrated honesty and anti-sycophancy for Claude and Codex agent workflows.

Stop Claude, ChatGPT, or any LLM from hedging, padding, narrating, and handing you homework. A battle-tested operating doctrine, not another "be concise" prompt.

A Cognitive Immune System for your Second Brain. Interrupts LLM sycophancy, forces objective pushback, and prevents delusional spiraling. v4.4.0: Now with Smart Routing.

Skills that teach an orchestrator bot to design one-job Grok Bot teammates, check each one before it ships, and write memory in the right tier. Cursor plugin with 12 Candor personas included. MIT.

Receipts, not vibes. MIT-licensed agent skill enforcing evidence-first AI output: conclusion first, facts vs unknowns, actions with owner/deadline/deliverable, no AI-slop, ADHD-friendly. Claude Code · Codex · Cursor. 拿证据来。

A Socratic, anti-summary reading companion for Claude Code — it argues with you, not for you.

A skill for direct, evidence-first answers without flattery, hedging, or false balance.

An experimental Reflective Intelligence architecture for epistemic clarity, self-understanding, and autonomous human choice. Not a chatbot. Not a guru. An instrument. Enforced seven-label epistemic system + runtime safety constitution.