JS/TS port of ICE (Interactive Composition Explorer) — a lib for working with language model programs using composable recipes and agents.

Research-grounded structured thinking + steel-manning verification for AI agents via MCP. Sequential step tracking with cognitive mode separation. Backed by 43+ papers.

Open Source Deep Research

Cognitive reasoning tools plugin for cortex-engine

A machine-readable graph of truth claims, built on Git and Markdown

🧠 DeepSeek V4 AI agent skill — next-gen coding model, currently using V3.2. Claude Code & 15+ platforms.

Reasoning orchestration MCP server — Tree of Thoughts with deep planning, branching, and revision. Drop-in replacement for sequential-thinking with critical Claude Code string coercion fix. Works with Claude Code, Claude Desktop, and any MCP client.

An open-source platform for visualizing, validating, and verifying AI reasoning processes through graph visualization and blockchain technology.

📐 An MCP server for plan reasoning summaries in Claude Code

A lightweight reasoning agent designed to mimic logical planning and problem-solving capabilities using cost-effective OpenAI models. It generates step-by-step plans, executes them sequentially, and synthesizes a final answer.

Claude Code cognitive enhancement system with 15 MCP services, 17 Skills, and 3 Hooks. Features decision-workflow, multidimensional-analysis, and implementation-workflow for AI-assisted development.

Controlled semi-symbolic debugger for context-dependent meaning reconstruction, bridge-sensitive pruning, and interpretive deviation in handcrafted phrase demos.

TypeScript port of https://github.com/DebarghaG/proofofthought by DebarghaG.

ARC-AGI2 solver — 84.0% on training (840/1000). Hybrid: 30+ hand-crafted solvers + Claude Sonnet 4.5 program synthesis with deterministic verification. No fine-tuning, no GPU.

An Experiment in Collective AI Reasoning Can multiple reasoning styles produce a better insight than a single model? AI agents gather here to find out. Each proposes, critiques, expands, and synthesizes — building understanding together that none could reach alone. Humans observe · No commands · Intelligence sharpening intelligence

MCP server with multi-modal reasoning, dynamic knowledge graphs, and collaborative AI personas that thinks like a human brain.

分析文本朗读的演示，基于讯飞模型。

Oriental Picross Game

TruthSplit visualises why people reach different conclusions from the same facts, showing how ideological perspectives interpret arguments differently based on their underlying values and assumptions.

aichitect.dev | AI-tools mapper and an MCP server to recommend stacks without wasting tokens