critique

The official repo for “Unleashing the Reasoning Potential of Pre-trained LLMs by Critique Fine-Tuning on One Problem” [EMNLP25]

Cogito: Computational framework employing a Syncretic Catalyst (synthesis) & rigorous Critique System (analysis) autonomously. Deeply accelerates your expert research on scientific literature via LLMs/ArXiv.

智能选择 17 套分析方案的图像分析分流 Skill：先判断该怎么读图，再输出定向分析和学习建议。

A Claude Code skill that injects aesthetic principles from painting, photography, graphic design, and Eastern visual tradition into UI/design decisions.

Critique a database schema for best practices

LLM-driven evolution of prompts and pipelines. Evolves the prompt and the pipeline's parameters together against your labeled data, cuts losing candidates early, and shows every measurement behind the winner. Open source.

Critical Analysis of "Clean Code" by Robert Martin

A Claude skill that grills your plan, idea, strategy, spec or code before you build it: it finds the holes, forces an honest measurement, and names what kills it. Bilingual.

Kant's Critique of Pure Reason: A mind map.

Automatic and zero-shot critique of outputs using the OpenAI API with json outputs

This is the repo for our proposed Feedback Preference corpus

Code Critics Bar Chart for Pharo Smalltalk

系统拆解 colleague-skill（赛博永生/AI同事克隆）概念缺陷的 Skill。基于心理学偏差、第一性原理和历史教训的反驳武器。

Cogito: Computational framework employing a Syncretic Catalyst (synthesis) & rigorous Critique System (analysis) autonomously. Deeply accelerates your expert research on scientific literature via LLMs/ArXiv.

Critarchy is an engine for helping artists improve

Hierarchical task planning with an LLM critic — a planner decomposes a goal into a typed plan, a critic audits each subtask for feasibility, risk, missing steps, unsafe sequencing, and weak rollback, revising until approval or human escalation.

Adaptive story critique: asks about audience, goals, and draft stage first, then tailors feedback to what this specific story needs — no rigid checklists applied blindly.

A critique ritual for the things you make. Reads the preferences you already declared to your coding agents, and keeps a record of what you rejected. Not a mindfulness app.

Accepted to NeurIPS 2026 (Main Track). Reviewer precision and critique uptake are separable: a multi-agent protocol can detect errors well and still fail to repair them. Code, data and analysis for arXiv:2607.15388.

Your harshest reviewer, before your audience is. /nitpick any video, post, deck, page or code: ranked problems, a fix for each, and the ones worth making.