d-wwei

Cognitive Base Creator — AI skill for Claude Code

AI community

A meta-skill for AI agents: generate cognitive base skills from any thinking framework.

How to install Cognitive Base Creator

This entry records only its repository, not the path inside it, so there is no exact command to give. Open d-wwei/cognitive-base-creator and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Cognitive Base Creator does

A meta-skill for AI agents: generate cognitive base skills from any thinking framework. Works with Claude Code, Codex, Gemini, Cursor, and any LLM agent.

Alternatives in AI

  • Karpathy LLM Wiki — Agent Skills-compatible LLM wiki for Claude Code, Cursor, and Codex 618 ★
  • Metaharness — 🛠️ The meta-harness for AI agents — scaffold your own focused, branded agent harness with its own npx CLI, MC 616 ★
  • Snip — CLI proxy that reduces LLM token usage by 60-90% 425 ★

README

Cognitive Base Creator

A meta-skill for AI agents: give it any thinking framework, and it generates a complete **cognitive base skill** — ready to install on Claude Code, Codex, Gemini, Cursor, or any LLM-based agent.

**[中文说明见下方](#中文说明)**


What problem does this solve?

The AI agent skill ecosystem is booming — design skills, coding skills, writing skills. But they all solve one problem: **what** to do. Nobody is solving the other problem: **how** to think.

A cognitive base skill is a different category of skill. It doesn't teach the agent to do a specific task — it changes the agent's reasoning process, improving the quality of **all** tasks.

Domain Skill Cognitive Base Skill
Analogy A recipe Knife technique
Layer Operational (what to do) Meta-cognitive (how to think)
Scope One task type All tasks
Effect Directly determines output Changes reasoning, indirectly improves output

This Creator automates the generation of cognitive base skills from any thinking framework.

How it works

Give the Creator a framework name:

"Generate a cognitive base from first principles thinking"

It produces a complete, installable package:

first-principles/
├── README.md                  ← Project documentation
├── cognitive-protocol.md      ← Core rules (~30 lines, always-on)
├── SKILL.md                   ← Full framework (~120 lines, reference)
├── anti-patterns.md           ← Framework-specific failure modes
├── examples.md                ← 3 before/after comparisons
└── install/
    ├── claude-code.md         ← Claude Code installation
    ├── codex.md               ← Codex installation
    ├── gemini.md              ← Gemini installation
    └── generic.md             ← Any other agent

Generation pipeline

  1. Framework Analysis — Extract core principles, map cognitive shifts, identify characteristic failure modes
  2. File Generation — Produce all 6 files following strict quality specs 3