Nimo1987

Nimo1987 / atomic-knowledge

Public

Markdown-first work-memory protocol for existing agents, with maintained knowledge, candidate notes, evals, and an example KB.

agent ai filesystem knowledge-base llm
30
0
100% credibility
Found Apr 08, 2026 at 30 stars -- GitGems finds repos before they trend. Get early access to the next one.
Sign Up Free
AI Analysis
Shell
AI Summary

Atomic Knowledge provides a markdown file structure and protocol for AI agents to maintain persistent, agent-editable work memory across sessions.

How It Works

1
📰 Discover Atomic Knowledge

You hear about a simple way to help your AI assistant remember research notes, links, and ideas across different chats without starting over each time.

2
📁 Set up your personal notebook

You create a special folder on your desktop that acts like a shared notebook for you and your AI to store and build knowledge together.

3
🔗 Introduce it to your AI

You share a guide note with your AI helper so it knows exactly how to read from and add to your notebook whenever you chat.

4
💡 Save your first ideas

Tell your AI to capture a link, note, or summary from your work, and it neatly organizes it into the notebook for future use.

5
❓ Ask questions anytime

In any chat, ask about past projects or comparisons, and your AI pulls the right details from the notebook to give smart, connected answers.

6
🧹 Keep the notebook fresh

Every now and then, ask your AI to review and tidy up the notebook, removing old stuff and highlighting what's important.

🎉 Your knowledge grows forever

Over weeks and months, your AI builds a rich, reliable memory of your research, making every session smarter and more productive.

Sign up to see the full architecture

5 more

Sign Up Free

Star Growth

See how this repo grew from 30 to 30 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 atomic-knowledge?

Atomic-knowledge is a markdown-first protocol that lets existing agents build and maintain a persistent work-memory layer, storing atomic units of knowledge like notes, sources, and insights across sessions. Instead of losing research context to chat resets, you init an atomic knowledge base with a simple shell script, plug the generated instructions into your agent, and it handles ingest, queries, writebacks, candidate notes, and maintenance—all in plain markdown files. It draws from Karpathy's LLM wiki idea, focusing on durable atomic knowledge proofs for long-running projects.

Why is it gaining traction?

It stands out by plugging straight into your current agent setup without new apps, dashboards, or vector DBs— just local files and shell commands for a portable atomic knowledge graph. Developers dig the cross-session continuity: agents proactively read entry pages, cite sources, and suggest upkeep like linting candidates or promoting notes, beating chat-history kludges or generic RAG rediscovery. The example KB and evals make integration testing dead simple.

Who should use this?

AI engineers running daily agent workflows for research threads, tool comparisons, or project synthesis, especially if session resets kill your momentum. It's for folks comfy with local markdown and shell who want maintained knowledge over persona memory or SaaS tools—think devs evaluating agent memory boundaries or building atomic structure knowledge organisers.

Verdict

Early release with 30 stars and 1.0% credibility score signals low maturity, but solid docs, example KB, evals, and health-check script make it easy to prototype. Worth a quick spin if you need agent work memory now; skip if you want battle-tested scale.

(198 words)

Sign up to read the full AI review Sign Up Free

Similar repos coming soon.