Windows SmartScreen warning?
ArcBrain doesn't yet have a code-signing certificate, so Windows Defender may show a warning when you run the installer. The file is safe — here's how to proceed:
Click "More info" in the warning dialog
Click "Run anyway"
macOS & Linux coming soon
Thanks for applying to the ArcBrain beta. Your info was sent and we'll follow up by email if you're selected.
Accepting 5 beta testers
The first 5 beta testers who give real feedback, bug reports, and product ideas will receive a free full ArcBrain Lifetime license.
What you getFree full lifetime access once you complete the beta and share useful feedback.
What we needHonest install notes, bugs you hit, rough edges, and new ideas that would make ArcBrain stronger.
Best fitPeople actively using VS Code, Cursor, Claude Desktop, or Windsurf who can test quickly and communicate clearly.
1
Open PowerShell as Administrator
Windows 10 / 11
irm https://arcbrain.pages.dev/install.ps1 | iex
This installs Python (if missing), clones ArcBrain, creates a virtual environment, installs dependencies, and opens the Setup GUI at localhost:9876.
Python 3.11+, pip packages, and Node.js (for the GUI) — only what's missing.
Step 2
Clone & configure
Pulls the latest ArcBrain from GitHub into ~/arcbrain and creates a virtual environment.
Step 3
Open Setup GUI
Opens localhost:9876 — connect Cursor, VS Code, Claude Desktop, Windsurf with one click.
Step 4
Start building
ArcBrain starts storing your AI sessions automatically. Open the Memory Dashboard any time.
System requirements
Requirement
Minimum
Notes
Operating System
Windows 10 / 11 (macOS & Linux coming soon)
Any modern OS
Python
3.11+
Installer adds it if missing
Node.js
18+
Installer adds it if missing
RAM
4 GB
8 GB recommended with Ollama
Disk space
500 MB
For models + memory database
Internet
Install-time only
Runs 100% locally after install
Ollama (optional)
Any version
Use OpenAI/Claude instead if preferred
Download FAQ
Is my data sent to the cloud?
Never. All memory — sessions, facts, code snippets, notes — is stored in a local SQLite database at ~/arcbrain/data/memory.db. Nothing leaves your machine unless you explicitly choose a cloud AI provider.
How do I update ArcBrain later?
Open the Setup GUI at localhost:9876 and click Check for updates. It runs git pull and pip install -r requirements.txt. Your memory database is never touched by updates.
Do I need a license to use ArcBrain?
All plans include a 14-day free trial with no credit card required. After the trial, choose Pro Monthly ($25/mo) or Pro Lifetime ($199) to continue. Your local memory data is always yours to keep regardless of plan status.
Can I use it without Ollama?
Yes. The memory features work without any AI model. Ollama (or OpenAI/Claude/Gemini/Grok) is only used for end-of-session summarization, which is optional.
What editors are supported?
VS Code (GitHub Copilot), Cursor, Claude Desktop, Windsurf, and Claude Code CLI. The Setup GUI connects all of them with one click.
Can I uninstall it completely?
Yes. Delete the ~/arcbrain folder and remove the MCP config entries from your editors (the GUI has a one-click Disconnect button for each). Nothing else is installed system-wide.
How does ArcBrain compare to Cursor's built-in memory?
Cursor's memory relies on .cursor/rules files and system prompts that you write and maintain by hand — static, session-scoped, and siloed to one editor. ArcBrain automatically captures decisions, errors solved, configs applied, and project context across all your editors (VS Code Copilot, Cursor, Claude Desktop, Windsurf) in a single persistent knowledge base. Memories survive every conversation, are semantically searchable, and follow you across projects and machines.
Is this related to Giga Memory?
ArcBrain is the self-hosted evolution of the Gigaboost / Giga Memory concept — built by ARC Technologies. Instead of a $70/month cloud service, ArcBrain runs entirely on your own machine: local database, local vector search, optional local AI via Ollama. Same persistent-context idea, zero subscription lock-in, and your data never leaves your computer.
Are there any API costs?
Only if you choose a cloud AI provider. If you use Ollama (local), there are zero API costs — everything runs free on your machine. If you connect a cloud provider (OpenAI, Claude, Gemini, Grok) for summarization and embeddings, expect roughly $1–$8/month in API usage depending on how active your sessions are. You can keep costs at $0 by using Ollama exclusively.
That said, cloud providers produce significantly better results — higher-quality session summaries, more accurate embeddings, and smarter memory recall. If you want the best experience, a cloud provider is worth the small cost.