Skip to content
hanfang/claude-memory-skill logo

hanfang/claude-memory-skill

Verified

Simple markdown-based memory for Claude agents without databases or embeddings.

SkillsAgent Skills 32Open source
View on GitHub
Updated 2026-06-16
hanfang/claude-memory-skill GitHub repository

What is hanfang/claude-memory-skill?

claude-memory-skill gives Claude a lightweight, file-based memory system. Core summaries live in core.md while detailed notes are stored in topic-specific files. The approach avoids any external services and keeps everything git-trackable and human-readable.

A main agent focuses on the user request and spawns a separate memory agent when needed. The memory agent loads relevant files, appends new entries with timestamps, and updates summaries. Session startup automatically loads core and profile information in the background.

It suits developers who want transparent, controllable memory for long-running Claude sessions. Users who prefer explicit commands over opaque built-in features will find the categorized structure and manual inspection tools helpful.

What you can build with hanfang/claude-memory-skill

Maintain coding preferences across sessions

Tell Claude to remember tool choices like pnpm over npm or specific auth requirements, and the skill stores them under the right topic for later recall.

Track debugging patterns

Log recurring issues such as async race conditions or timeout values so the agent can surface past fixes when similar problems appear.

Build a personal profile

Edit me.md once with your background and preferences so every new session starts with context about your role and tech stack.

Install hanfang/claude-memory-skill

Install
curl -fsSL https://raw.githubusercontent.com/hanfang/claude-memory-skill/main/install.sh | bash
Quick start
curl -fsSL https://raw.githubusercontent.com/hanfang/claude-memory-skill/main/install.sh | bash
  1. 1Run the install script via curl or clone the repo and execute install.sh locally.
  2. 2Confirm that CLAUDE.md and the memory directory are created under ~/.claude.
  3. 3Open me.md and add facts about yourself and your workflow.
  4. 4Start a new Claude session and let it automatically load core memory in the background.
  5. 5Use natural language like 'remember that...' to trigger saves or run /mem commands for manual checks.

hanfang/claude-memory-skill: pros & cons

Pros

  • +No external dependencies or vector stores required
  • +Memory operations run in the background without blocking the main agent
  • +Files are plain markdown so they stay readable and version-controllable
  • +Hierarchical structure keeps summaries quick to load while details stay available on demand

Cons

  • Requires Claude Code environment and the specific skill hook
  • Lacks any semantic search so retrieval depends on topic names and grep
  • Manual topic management can become tedious for very large histories
Did you find this helpful?

Frequently asked questions

No, the skill relies on Claude Code's command hooks and background agent support.

User reviews

Verified reviews from the community shape this listing's rating.

Loading reviews…

Sign in to review

Promote hanfang/claude-memory-skill

Add this badge to your website, or share the tool.

DFeatured on Dhanasvihanfang/claude-memory-skill 0