Anything you'd hand a senior sysadmin.
From CVE exposure reports to deploy receipts, every agent output is actionable and auditable.
Powered by leading AI infra
Ask in plain language. Get answers with charts, tables, and actions — every response is an artifact you can act on.
Models failing policy checks are mostly outdated versions without signed attestations. 8 can be auto-updated; 4 require manual review due to custom fine-tuning.
Local, Docker, or cloud — same stack, different runtime.
Auto-detects GPU (4090/Z790), installs drivers, CUDA, Docker, and the full stack. Production-ready systemd service.
git clone https://github.com/ProjectZeroDays/FreeAI_AI_Inference_Workstation.git
cd FreeAI_AI_Inference_Workstation
sudo ./hardware/install-stack.sh
bash models/auto-download-models.shDashboard: localhost:8030 · Agent API: localhost:8050
Any host with NVIDIA Docker. Split or all-in-one profiles. Zero host dependency management.
git clone https://github.com/ProjectZeroDays/FreeAI_AI_Inference_Workstation.git
cd FreeAI_AI_Inference_Workstation
docker compose --profile allinone up -d
# Open http://localhost:8030Profiles: allinone, gpu, cpu, agents
Production-grade with GPU operator, autoscaling, and service mesh. Full manifest in k8s/freeai-deployment.yaml.
kubectl apply -f k8s/freeai-deployment.yaml
# Services: 8030 (dashboard), 8050 (agent API)GPU operator · HPA · PVC for models
Salad, Vast.ai, RunPod, Hetzner, Hostinger, DigitalOcean. Same stack, pre-configured templates.
docker pull ghcr.io/freeai/freeai:latest
# or use cloud provider templatesFrom $0.20/hr on Salad GPU clusters
FreeAIOS — a bootable Live ISO with FreeAI tools pre-installed. Try it before you commit.
Default variant. Full XFCE desktop, FreeAI stack pre-loaded, NVIDIA driver ready. Best for general use.
Penetration testing focus. Kali tools + FreeAI agents for security assessment on any target machine.
Privacy-first distro. Encrypted workspace, anonymous browsing, FreeAI security agents pre-configured.
Reproducible infrastructure. Declarative config, atomic rolls back, ideal for CI/CD environments.
# On an Ubuntu host:
sudo apt-get install -y xorriso isolinux
UBUNTU_ISO=ubuntu-24.04.2-live-server-amd64.iso ./live/build-live.sh
# Optional: bake repo into ISO for offline install
REPO_TARBALL=../dist/unified-ai-stack-v1.2.0.tar.gz ./live/build-live.sh
# Result: freeaios-amd64.iso — boot from USBBoot menu: Try Live · Install to disk · Rescue shell · Memory test
Linux, macOS, and Windows — each with its recommended deployment path.
Native support. Ubuntu, Debian, Kali, Kodachi, NixOS. GPU acceleration via CUDA / ROCm.
sudo ./hardware/install-stack.sh:8030Via Docker Desktop or Colima. CPU mode natively; GPU offload to Linux hosts or cloud GPUs.
docker compose --profile allinone up -d:8030via WSL2 + Docker Desktop, or native Docker with NVIDIA drivers. Full GPU passthrough in WSL2.
docker compose --profile allinone up -dReal-time telemetry, log streams, skill management, and the FreeToken desktop console — all in one interface.
3-column scroll-spy layout with filterable page TOC, live section highlighting, and AI chatbot for repo-only Q&A.
15 simulated services with live log injection every 3s. Text/JSON export, pattern search, and AI help modal.
12 telemetry cards — upload/download speeds, VPN/Tor/DNSCrypt status, IPv4/IPv6. Drag-drop reorder, 5s live refresh.
Draggable web cards for edge runtime config, skill/MCP/plugin links, telemetry insights, and per-card recommendations.
Community-driven content. JSON-backed posts and threads with dark/light theme matching the dashboard.
22 providers, 10 clients, cross-installable skills/MCPs/plugins/agents. One config file, all clients updated.
From quantum C2 to creative generation, from shell scripting to music synthesis — skills are peer-reviewed, versioned, and shareable.
APT hunting, exploit research, social engineering, LotL techniques, evasion, payload generation.
Hardening, compliance (NIST/FedRAMP/DOD), SIEM integration, incident response, forensic analysis.
Red↔blue orchestration, control validation, remediation bridging, autonomous validation.
ASCII art, architecture diagrams, pixel art, manim videos, ComfyUI, p5.js, songwriting.
CI/CD, Docker, Kubernetes, Terraform, monitoring, Kanban orchestration, webhook subscriptions.
Model serving (vLLM, llama.cpp), fine-tuning (SFT/DPO/RFT), evaluation harness, RAG pipelines.
Plugin system with 6+ core plugins and a marketplace. Zero-downtime updates, version-pinned, and sandboxed.
GitHub Actions, Docker publish, release bundling — triggered on push or schedule. Zero-downtime deployments.
File-system watcher for config changes, model downloads, and RAG document ingestion.
Config snapshot, registry export, weekly retention (keep 10). One-command restore.
Supervisor loop (10s), health agent (30s), recovery agent (15s). systemd units with auto-restart.
Browse, install, and share plugins. Version-pinned installs with rollback support.
Cron-driven daily cleanup, weekly backup, timed idle windows with auto-restore on activity.
Pre-configured servers for memory, code intelligence, browser automation, search, LLM access, and more — all wired into the router.
Custom extensions for VS Code, Cursor, and CLI — unified config, shared skills, and one-click install from the dashboard.
SKILL.md editor, skill catalog browser, pattern monitor — all from within Cursor.
Project navigation, model selection, agent runner — open any FreeAI project instantly.
freeai-cli with 14 subcommands: providers, test, deploy, skills, mcp, agents, workflow.
Native Electron app with tray icon, auto-update, and system-wide model routing.
Highlight any text, send to FreeAI for analysis, save as skill snippet.
iOS and Android with full feature parity — monitor agents, trigger workflows, view logs.
Visual workflow designer with validation, templates, audit logs, and export/import. From one-shot prompts to scheduled automation.
Drag-and-drop pipeline builder. Sequential + parallel steps, conditional branching, error handling.
4 shipped templates: code review, deployment, security scan, RAG ingestion. Create your own.
JSONL audit trail for every run. Re-run, replay, and export pipeline history.
plan → code → verify (real compilers/tests) → fix → review → document → package. One spec, full project.
Semgrep → Bandit → Safety → Trivy → auto-patch → verify. CI-ready with pass/fail gates.
Scheduled runs: daily reports, nightly backups, weekly audits, morning briefings. Delivery to any platform.
Built on what FreeAI delivers — a unified stack that grows with you.
Every model state understood in context — latency, cost, capacity, and fallback chains across 22 providers.
24 agents deployed, updated, and retired by policy — across every framework and language.
Security embedded in daily operations — not bolted on afterwards. Aikido + pentest + auto-patch.
Infra that feels invisible to the people building on it. One dashboard, one config, one stack.
Three steps to a fully autonomous AI workstation.
Runs on the stack you already use. Clone, install deps, and launch. No new infrastructure required.
Providers, agents, skills, MCPs — modeled deeply enough to predict fallbacks, costs, and capacity.
Approvals, rate limits, full history. Autonomy you can audit. Auto-restart on failure built in.
Salad GPU rental, Vast.ai instances, or bare-metal. Same stack, different compute.
Local, Docker, or cloud — same stack, different runtime. View all deploy options →
git clone https://github.com/ProjectZeroDays/FreeAI_AI-Inference-Workstation.git
cd FreeAI_AI-Inference-Workstation
sudo ./hardware/install-stack.sh
bash models/auto-download-models.shDrivers → CUDA → Docker → stack → systemd → UFW. Auto-detects GPU and configures everything.
git clone https://github.com/ProjectZeroDays/FreeAI_AI-Inference-Workstation.git
cd FreeAI_AI-Inference-Workstation
docker compose --profile allinone up -d
# Open http://localhost:8030Any host with NVIDIA Docker. Split or all-in-one profiles.
Kubernetes manifests, Vast.ai templates, RunPod kits, Hetzner bare-metal. Also bootable as a Live ISO — Ubuntu/Kodachi/Kali/NixOS with install, live, and rescue modes.
docker pull ghcr.io/freeai/freeai:latest
# or boot freeaios-amd64.isoDescribe what you need in plain language — FreeAI builder agents scaffold the complete project with code, tests, and docs.
FastAPI + React, Django + Next.js, Laravel + Vue with database and auth.
Landing pages, portfolios, SaaS sites with Next.js + Tailwind.
Custom CRM with contacts, deals, tasks, and email integration.
FAQ bots, ticketing, knowledge-base powered support.
Booking bots for salons, clinics, auto shops, spas, and more.
Lead capture, CRM sync, email sequences, conversion tracking.
Multi-channel campaigns with analytics and A/B testing.
Google/Facebook/LinkedIn ads with budget tracking and ROAS.
200+ platform actions exposed to AI. Everything Intelligence can see, it can do — with your permission.
Agent-to-agent orchestration across systems — FreeAI agents cooperate with the rest of your stack.
Professional-grade security scanning, vulnerability management, and compliance tracking built in from day one.
Scan code, test apps, auto-patch from the dashboard. Real-time compliance scoring.
Semgrep, Bandit, Safety, Trivy — comprehensive SAST/DAST scanning.
Generate and apply safe fixes for critical and high vulnerabilities.
Monitor CVEs, auto-update with presets (conservative/balanced/aggressive/secure).
14 Red Team, 12 Blue Team, 7 Purple Team expert skills for every vector.
Up to 10 keys per provider, round-robin/random/least-used, auto-pause on 429.
No — use MOCK_LLM=1 for CPU-only mode, or connect to hosted APIs (OpenAI, Anthropic, etc.) via the provider system.
22+ hosted APIs including OpenAI, Anthropic, Gemini, Groq, Mistral, DeepSeek, and more via the provider adapter system.
Yes — FreeAI is designed for production with auth, rate limiting, health checks, and auto-restart policies.
Yes — MIT licensed. Full source at GitHub.
Use the Skills API or add SKILL.md files to the skills/ directory. See the docs for details.
Yes — any OpenAI-compatible endpoint works. Configure in config/providers.json or via the dashboard.
Deploy FreeAI in minutes. Local LLMs, autonomous agents, full SDLC automation — on your hardware, your terms.
Fully integrated · works with the stack you already run
v1.2.0 — Autonomous SDLC Agents
FreeAI agents plan, code, test, and ship full projects — from a single prompt.
Fully integrated — works with the stack you already run