Skip to main content
The Mission

Self-Hosted AI.
Full Control. Zero Leakage.

Enclavetools is the curated directory of private AI tools that run on your own hardware. No API keys. No surveillance. No vendor lock-in.

The Hidden Cost

Every "Free" AI Service
Has a Hidden Cost

Every time you use an AI API, you're trading something more valuable than money: your data, your independence, and your security.

Data Leakage

Your prompts, code, and proprietary documents train the next model.

Vendor Lock-In

Pricing changes overnight. Models deprecate. Features disappear.

Rate Limits

You're capped, throttled, and queued behind paying customers.

Privacy Theater

"Enterprise security" still means your data touches someone else's server.

The AI industry built a beautiful prison. The walls are convenience. The guards are API keys. And you're paying for the privilege of being locked in.

It doesn't have to be this way.

The Solution

Your AI Stack.
Your Hardware. Your Rules.

Enclavetools is the definitive directory of self-hosted AI tools — software that runs entirely on your own infrastructure. Every tool listed here is screened by coding agent, and verified by a human before it goes live.

1

Runs Locally

On your laptop, your server, or your Raspberry Pi cluster. No cloud required.

2

No API Keys

Zero monthly bills. Zero per-token pricing. Pay for your hardware once.

3

Data Stays Yours

Zero telemetry. Zero outbound data leakage. Verified by packet-sniffer audits.

4

Yours Forever

No deprecation. No sunset. No enshittification. The software can't be taken from you.

"It was 3 AM during a production outage when I dug into our network traffic and realized just how much sensitive user data our closed-source AI vendors were scraping off our calls. That night, I stopped trusting closed models. I started building what I wished existed: a directory of privacy-first, local-capable AI tools that answer to no one but you."
Mamy Rakotomalala

The Founder

Mamy Rakotomalala

Infrastructure engineer and AI privacy advocate. After years of building ML systems on cloud APIs, I dug into our network traffic during a production outage and discovered how much sensitive user data closed-source AI vendors silently scrape. That moment changed everything.

I started Enclavetools to build a directory of AI tools that answer to no one but you. Every tool is audited — not because we trust code, but because we don't trust anything we haven't verified.

The Directory

Automated Static Discovery,
Human Verification, Deep audits on demand.

A coding agent statically reviews the source of every candidate — dependencies, telemetry SDKs, outbound-call patterns, license terms — before it's shortlisted. A human resolves the edge cases before anything goes live.

Listed & Verified

Every tool, by default

What a code-level read can confirm, for all 163+ tools, at no cost to you.

  • License & commercial-use check
  • Dependency & known-tracker scan
  • Docs claims vs. code diff
  • Declared hardware requirements
◆ Deep Audited

On request, per tool

The tool is actually installed and run, with real traffic captured — for teams who need certainty, not just a first read.

  • Full runtime network capture
  • Every endpoint classified & documented
  • Measured benchmarks on real hardware
  • Report published back to the tool page
🔍

Plain-English Overview

What the tool actually does, who it's for, and when you'd use it — not marketing copy.

📋

Technical Scorecard

License safety, commercial use, API support, fine-tuning, quantization, Docker, GUI, telemetry, and offline capability — scored and compared.

⚙️

Hardware & Setup

RAM, GPU, architecture, and setup difficulty — from Plug & Play to Advanced — so you know before you clone.

The Methodology

4 Stages of
Due Diligence.

Every tool passes through the same four-stage pipeline before it's published. No listing goes live without evidence.

Stage 1

Discover + Scan

Mining GitHub Topics, HuggingFace Hub, and community sources for candidates. Automated license scan via ScanCode + FOSSology.

GitHub API HuggingFace ScanCode FOSSology
Stage 2

Static Code Review

A coding agent reads the source tree via the GitHub API: dependency tree, known telemetry/analytics SDKs, outbound-call patterns, and how the code's behavior compares to what the docs claim. Unresolved cases are flagged for human review.

Static Analysis Dependency Tree Claim vs. Code
Stage 3

Declared Requirements Check

Confirm stated RAM/VRAM, OS support, and hardware targets from the repo and docs. Check commit activity and release cadence via pip-audit and npm audit for known vulnerabilities.

Health Signals pip/npm Audit
Stage 4

Review + Publish

Dependency supply chain audit via NVD CVE database. Edge cases, ambiguous licenses, unknown network calls — resolved by a human before going live.

Dependency Scan CVE Check Human Review

Go Deeper

Deep Audits & Sizing Reports

Every tool clears the standard pipeline before it's listed. Some — the ones entering production or compliance-sensitive environments — deserve more than a code read.

🔬

Deep Network Audit

Extended runtime observability with eBPF-based network tracing across real usage sessions. Every outbound endpoint, connection, and payload — captured at the kernel level with zero interception, identified, classified, and documented in a report you keep.

📐

Benchmark & Hardware Sizing

Real throughput numbers — tokens/sec, latency, memory footprint — on defined hardware profiles, plus a sizing recommendation for your actual workload.

Open the Calculator

Why It Matters

The Future of AI
Is Personal

The AI industry is consolidating into a handful of walled gardens. But the most powerful AI applications won't run in someone else's data center — they'll run on your devices, in your context, with your data.

🛡️

Sovereignty

Your infrastructure, your rules. No external dependencies. No third-party can pull the plug.

💰

Economics

Pay for hardware once. No recurring API bills. The economics shift from OpEx to CapEx.

🚀

Innovation

Experiment freely. Fine-tune. Break things. Learn. No approval process, no billing alerts.

Resilience

Your tools work offline. No cloud outage, no API downtime can stop your workflow.

The Vision

A World Where Every Business and Developer
Runs Their Own AI

Not replacing cloud AI — giving you a real choice. A choice not between providers, but between their infrastructure and yours.

Phase 1

Comprehensive Coverage

Every self-hosted AI tool worth knowing, across every category. If it runs on your hardware and respects your privacy, it belongs here.

Phase 2

Compatibility Mapping

Knowing exactly what works with what. Stack Builder already lets you compose compatible tools — we're mapping every integration path.

Phase 3

One-Click Deployments

Self-hosting shouldn't require a PhD. We're building toward a future where deploying your own AI stack is as easy as using an API.

Join the Mission

Start Building on Your Terms

This isn't a company. It's a mission. Browse the registry, find your stack, and take control of your AI infrastructure.

Know a tool we're missing? Send it in.