Run Cabinet AI Knowledge Base

🚀 Quick Verdict

Run Cabinet solves the context window problem by giving your local AI agents a permanent place to live and read files. It is refreshing to use a tool where data stays on your machine rather than a third-party cloud. The terminal integration works surprisingly well for running code directly from the browser interface.

Overall Score	8.8/10
Best For	Privacy-conscious founders
Tested Plan	Self-hosted Open Source
Testing Period	7 days
Biggest Strength	Local data persistence
Biggest Weakness	Initial command-line setup
Best Alternative	Obsidian

🤔 What Is Run Cabinet?

Run Cabinet is an open-source “Startup OS” designed to hold all your business context in one place. Unlike traditional wikis, it is built specifically for AI agents to interact with your files, allowing them to remember project history across different sessions. It was created by Hila Shmuel, a former Apple engineering manager, to fix the issue of context evaporation in AI chats.

The tool functions as a knowledge base where you can store markdown files, PDFs, and even embed full-screen HTML apps. By running locally via specialized agents, it allows you to automate repetitive business tasks like scouting Reddit or generating weekly reports without sending your proprietary data to a foreign server.

⚙️ How We Tested Run Cabinet

We installed the self-hosted version using the npm command on a local machine for a one-week trial. Our testing focused on three founder-level tasks: importing 20+ existing startup documents to test the markdown rendering, setting up a Scheduled Job to monitor specific topics, and using the Missions board to track a product launch. We specifically tested the Git-Backed History by making multiple edits to a single page to verify the auto-commit feature.

✨ Key Features (What Actually Stood Out)

The architecture is designed for founders who need high-speed productivity without the overhead of complex database management. Since it uses files on disk, your data remains portable and human-readable even if you stop using the app.

• Scheduled Jobs — We set a cron-based agent to scout for mentions of our brand; it successfully updated its own log file every 6 hours without manual intervention.
• Embedded HTML Apps — You can drop an index.html into a folder and it renders as a functional dashboard, which is excellent for custom internal calculators.
• Git-Backed History — Every save triggers an auto-commit, allowing us to see a full diff and restore old versions of our strategy docs instantly.
• Web Terminal — Testing commands directly in the browser using the xterm.js integration felt much smoother than constantly switching back to a separate terminal app.
• Missions & Tasks — The Kanban view for agents makes it easy to visualize which AI “team member” is working on specific goals like content editing or market research.

💰 Run Cabinet Pricing — Is It Worth It?

Currently, Run Cabinet is entirely free and open-source, making it a zero-risk addition to your automation stack. You provide your own API keys for models like Claude or OpenAI, meaning you only pay for the tokens you actually consume rather than a flat monthly SaaS fee.

Plan	Price	Best For	Watch Out For
Open Source	$0	Founders and Developers	Requires local installation

Our pick: Open Source — This is currently the only way to run the tool, and it offers the best privacy for sensitive business data.

🧪 What We Found During Testing

Setting up the tool was faster than expected, taking about three minutes from terminal input to seeing the UI. We found the File-Based Everything approach to be a massive advantage when we wanted to bulk-edit files using a different text editor; the changes reflected in Cabinet instantly. However, the AI agents occasionally struggled with very large PDF files if the local context window wasn’t configured correctly in the settings.

A founder in our community who runs a content agency told us he used Cabinet to keep his client brand voices consistent across different AI sessions without having to re-upload style guides every time.

⚠️ Limitations We Found

• Local Setup Required — As of early 2025, there is no hosted cloud version, so you must be comfortable using a terminal to get started.
• Early Stage Software — Some UI elements in the Missions board felt a bit thin, and we noticed occasional glitches when dragging files between deep folder structures.
• BYO-AI Keys — You must manage your own API billing with providers like Anthropic, which might be an extra step if you prefer all-in-one pricing.

⚔️ Run Cabinet vs Competitors

Founders often compare this tool to traditional note-taking apps that are slowly adding AI features.

Competitor	Pick it instead of Run Cabinet if…
Notion	You need a collaborative cloud workspace for a large human team.
Obsidian	You want a more mature plugin ecosystem for personal knowledge management.
NotebookLM	You only need to chat with a few specific PDFs and don’t care about automation.

👍 Pros & Cons

✅ Pros	❌ Cons
Full data privacy (local storage)	Requires npm/node to install
Automated AI agents with cron logs	No mobile app available yet
No monthly subscription fees	Early-stage UI bugs
Portability via markdown files	Manual API key management

🎯 Who Should Use Run Cabinet (And Who Shouldn’t)

✅ Use it if you:

• Handle sensitive client data or intellectual property that shouldn’t leave your machine.
• Want to build a persistent memory bank for your AI agents to reference.
• Need a lightweight way to run scheduled AI tasks like web scraping or reporting.

❌ Skip it if you:

• Prefer a hosted solution where you don’t have to manage software updates — try Notion instead.
• Are not comfortable using a command line for the initial installation.
• Need advanced project management features like Gantt charts or time tracking.

❓ Frequently Asked Questions

Is Run Cabinet really free?
Yes, the core software is open-source and free, though you still pay for the AI tokens used by your API keys.

Where is my data stored?
Your data remains entirely on your local hard drive in a folder of markdown files you specify during setup.

Can I use local LLMs with Cabinet?
Yes, the tool is built on a Bring Your Own AI (BYOAI) principle and supports local models alongside providers like Claude.

Do I need to be a developer to use it?
While installation requires the command line, the daily usage is through a visual browser interface designed for founders.

What is the Karpathy Effect?
It refers to the compounding value of feeding structured context into an LLM, which Cabinet facilitates through its persistent storage.

Does it support collaboration?
Since it is local-first, collaboration is usually handled by syncing the Cabinet folder via Git or a shared drive.

Looking for more tools like this? See all productivity tools we’ve reviewed →