blader

blader / napkin

Public

A Claude Code skill that gives the agent persistent memory of its mistakes via a per-repo markdown scratchpad.

327
21
100% credibility
Found Feb 07, 2026 at 140 stars 2x -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
AI Summary

Napkin is a simple enhancement for AI coding assistants that builds long-term memory by tracking mistakes, corrections, and successes in a personal notebook within your project.

How It Works

1
💡 Discover Napkin

You hear about Napkin, a clever trick that helps your AI coding buddy remember mistakes from past work sessions.

2
📥 Add the helper

You quickly add Napkin to your AI coding setup in one easy step, and it's ready to go.

3
🚀 Kick off your first session

Your AI buddy starts a new coding adventure and creates a personal notebook to track wins and slip-ups.

4
✏️ Work side by side

As you build your project together, the AI notes down errors, your fixes, and handy tips right away.

5
🔄 Jump into more sessions

Each time you return, your AI reviews its notebook and picks up right where it left off, getting sharper.

6
✨ Notice the magic

After a few sessions, your AI stops repeating old mistakes and even spots problems before you do.

🎉 Smarter coding partner

Your AI helper has transformed into a reliable expert that saves you time and hassle every day.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 140 to 327 stars Sign Up Free
Repurpose This Repo

Repurpose is a Pro feature

Generate ready-to-use prompts for X threads, LinkedIn posts, blog posts, YouTube scripts, and more -- with full repo context baked in.

Unlock Repurpose
AI-Generated Review

What is napkin?

Napkin is a lightweight skill for Claude code agents that adds persistent memory by maintaining a markdown scratchpad in your repo. It logs the agent's mistakes, your corrections, successful patterns, and repo-specific quirks, reading them at session start and updating live during work. Install via a simple git clone to your Claude Code or Codex skills directory—activates automatically per repo for gradual behavior improvement over sessions.

Why is it gaining traction?

In a sea of Claude GitHub integrations like PR reviews and copilot-style tools, Napkin stands out with "baby continual learning" that makes agents smarter without complex setups or pricing tiers. Developers notice the shift by session 3-5: fewer repeated errors, proactive fixes, and personalized workflows. Its zero-config install and per-repo isolation hook tinkerers exploring Claude code changelog and docs on GitHub.

Who should use this?

Backend devs iterating on Claude GitHub actions in private repos, AI prompt engineers refining agent outputs over multiple sessions, or teams sharing learned patterns via committed scratchpads. Ideal for anyone using Claude code web or connectors tired of resetting context each time.

Verdict

Grab it if you're deep into Claude code GitHub flows—232 stars and solid README show early promise despite the 1.0% credibility score signaling niche maturity. Low risk at MIT license, but watch for broader testing as napkin AI evolves.

(178 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.