STING runs AI on your servers. Not ours. Not OpenAI's. Yours. Your data never leaves your infrastructure unless you explicitly tell it to.

We also built Honey Jars—encrypted knowledge containers you can search and share. Think of it as turning your institutional knowledge into something actually useful.

You can! And if you want to spend your weekends building PII detection, auth systems, and knowledge management from scratch... honestly, go for it. It's educational.

STING gives you all that out of the box:

• Honey Jars: Encrypted, searchable knowledge bases
• PII Protection: Sensitive data auto-detected and scrambled
• WebAuthn: Passwordless login with biometrics
• RAG Pipeline: Document processing and vector search
• One Install: Not six weekends of debugging

Yes. Apache 2.0 licensed. Open source. No credit card. No "trial period." No artificial limits.

Two ways to run it:

• Nectar (Desktop): Personal AI on your machine. Mac/Win/Linux.
• STING Server: Self-hosted for teams. Docker-based.

Both give you:

• Bee AI assistant
• Unlimited Honey Jars
• PII scrambling
• Local LLM support (Ollama)
• WebAuthn/Passkey auth

Fork it, modify it, run it forever. No vendor lock-in.

Encrypted knowledge container. Your docs go in, Bee AI can search them, but nobody sees the raw files unless you say so.

Quick features:

• AES-256 encryption
• AI can search without exposing content
• You control who accesses what
• Full export—your data is yours

Example: Legal puts 500 contracts in a Honey Jar. Sales asks "California NDA template?" Bee AI finds it. Sales gets their answer. Nobody sees the actual contract files. Legal doesn't get 47 Slack messages. Everyone wins.

The name: Bees collect nectar and store it in honeycomb. You collect knowledge and store it in Honey Jars. Our desktop app is called Nectar. It all connects.

Multiple layers. Not just marketing checkboxes.

Data Protection:

• Your data never leaves your infrastructure
• AES-256 encryption at rest
• TLS 1.3 in transit
• Honey Jars are encrypted

PII Protection:

• PII types auto-detected
• Scrambled before AI sees it
• Passkey auth for sensitive operations

Access Control:

• WebAuthn/Passkey authentication
• Role-based access control
• No passwords to steal

Security deep-dive →

Most of them, honestly.

Local (via Ollama):

• Qwen 2.5: Great all-rounder
• Phi-4: Good for reasoning
• DeepSeek Coder: Code generation
• LLaVA: Images + text
• Anything Ollama runs: Llama 3, Mistral, whatever

Cloud (optional): OpenAI, Anthropic, or any OpenAI-compatible API. Mix local and cloud based on your needs.

From paste-command-to-asking-questions? Yeah, about 30 minutes.

• Basic setup: ~30 minutes
• Production with your data: 2-5 days
• Full deployment with security review: 1-2 weeks

One command:

bash -c "$(curl -fsSL https://raw.githubusercontent.com/AlphaBytez/STING-CE/main/bootstrap.sh)"

It figures out your OS, checks requirements, launches a setup wizard, and starts everything. You answer some questions. It does the work.

Full install guide →

After install? Zero internet required.

• No phone-home
• No telemetry
• No sneaky API calls
• No license verification

Pull the ethernet cable. STING doesn't care. It runs local.

One catch: Initial install needs internet to download models and dependencies. After that, air-gap away.

Hybrid option: If you want, you can configure cloud AI for non-sensitive tasks. But it's opt-in, not required.

Nectar (Desktop):

• Mac, Windows, or Linux
• 8GB RAM (16GB recommended)
• 10GB storage (more for AI models)
• Docker and Ollama (installed automatically or manually)

STING Server:

• Ubuntu 20.04+, Debian 11+, macOS, or WSL2
• 8GB RAM (16GB+ recommended)
• 4 CPU cores (8+ recommended)
• 50GB storage (100GB+ SSD recommended)
• GPU optional but helps

Docker gets auto-installed if you don't have it.

Nectar: Desktop app for personal use. Mac/Win/Linux. Works offline with local LLMs. Single user. Can connect to a STING Server if you want.

STING Server: Self-hosted server for teams. Docker-based. Multi-user with WebAuthn auth. Runs on any Docker host.

Start with Nectar if you want to try STING quickly on your own machine. Deploy STING Server if you need multi-user access or want to share knowledge bases across a team.

Full comparison →

Yes. Nectar can connect to a STING Server, so you can:

1. Start with Nectar to evaluate locally
2. Deploy STING Server when your team is ready
3. Connect Nectar to your server for a native desktop experience with shared knowledge

Nectar (Desktop):

1. Download from GitHub Releases
2. Install Ollama if you haven't
3. Run the app

STING Server:

bash -c "$(curl -fsSL https://raw.githubusercontent.com/AlphaBytez/STING-CE/main/bootstrap.sh)"

It handles the rest.

Community support:

• GitHub Discussions
• GitHub Issues
• Discord
• Documentation

We're friendly. Ask questions. Report bugs. Suggest features.

Absolutely. STING is open source.

• Check out issues labeled good first issue
• Documentation improvements always welcome
• Bug fixes, feature PRs, whatever

GitHub repo →