mem-init

Initialize canonical project memory and generate AGENTS.md and CLAUDE.md wrappers.

10|1|Updated Feb 21, 2026
One-click install
npx skills add https://github.com/dongzhuoyao/tao-research-skills --skill mem-init
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mem-init
Source: https://github.com/dongzhuoyao/tao-research-skills/tree/main/mem-init
Command: npx skills add https://github.com/dongzhuoyao/tao-research-skills --skill mem-init

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) components.

What problem does it solve?

Repositories often lack a single canonical memory file and standardized AGENTS.md/CLAUDE.md wrappers, which leads to duplicated instructions, inconsistent agent behavior, and manual setup overhead for Codex and Claude Code workflows.

Core Features & Use Cases

  • Canonical Memory Bootstrap: Creates memory/project.md as the single source of truth for agent memory.
  • Safe Migration: Detects existing CLAUDE.md or AGENTS.md content and migrates non-generated content into memory/project.md to preserve prior instructions.
  • Wrapper Generation: Generates AGENTS.md from memory/project.md and creates a CLAUDE.md wrapper that references AGENTS.md, plus an optional memory/claude.md overlay and a starter README.md.
  • Integration: Delegates final initialization steps to the memory-sync module for consistent, repo-wide initialization.
  • Use Case: Quickly prepare a new or legacy repository for Codex and Claude Code agents by consolidating memory and generating host files.

Quick Start

Use the mem-init skill to initialize canonical project memory, migrate any existing AGENTS.md or CLAUDE.md content, and generate AGENTS.md and CLAUDE.md in the target repository.

Frequently Asked Questions about mem-init

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I initialize a canonical project memory file for Claude Code and Codex agents?

Initialize canonical project memory by creating a memory/project.md file as the single source of truth, then generating AGENTS.md and CLAUDE.md wrappers to provide unified instructions for Codex and Claude Code agents.

Can I migrate existing AGENTS.md or CLAUDE.md content into a unified memory file?

Yes, safe migration detects existing AGENTS.md or CLAUDE.md content and moves non-generated instructions into memory/project.md, preserving your prior agent configurations and preventing data loss during repository initialization.

What is the best way to set up a repository for both Codex and Claude Code workflows?

The best way to standardize Codex and Claude Code workflows is generating a CLAUDE.md wrapper that references AGENTS.md, adding an optional memory/claude.md overlay, and writing a starter README.md for consistent agent behavior.

Does initializing project memory overwrite my existing repository files?

Initializing project memory performs safe migration of existing files and supports an optional force overwrite flag. By default it protects non-generated content while delegating final repo-wide initialization to a memory-sync module.

Why do I need a canonical memory file for agent workflows?

You need a canonical memory file to eliminate duplicated instructions and inconsistent agent behavior across Codex and Claude Code, consolidating configurations into memory/project.md to reduce manual setup overhead.