Skip to main content

Kailash OS

A NixOS-based Linux distribution for AI security — one declaratively built environment for offensive, defensive, and platform tooling

License: BSD-3-Clause release OpenSSF Scorecard OpenSSF Best Practices

Kailash OS keeps the distribution contract that made Kali Linux the default security environment: curated tooling, category navigation, live images, and self-consistent surfaces. The AI systems stack is the surface Kali Linux never mapped. Kailash OS applies the same contract to this stack. The Nix flakes define everything that attacks, defends, or runs an AI system exactly once.

The four pillars

Kailash targets four domains. Each pillar ships twice: offensive tooling in the ATTACK layer; build-side or defensive tooling in the DEFENCE and OPS layers:

  • AppSec — classic application security assessment (CLASSIC layer, software only)
  • APISec — web, API, and AI-endpoint runtime assessment
  • MLSec — adversarial ML: evasion, poisoning, extraction, and inference attacks (predictive AI)
  • AISec — LLM, agentic, and MCP security: prompt injection, tool-use attacks, supply chain

Taxonomy

The tool set carries a CORE substrate plus four named tool layers. The taxonomy holds 32 census categories: 31 shippable categories + 1 opt-in data pack. CORE — kernel, base userland, hardening, safety defaults, desktop, CLI and menu engines — hosts no tool categories; every image inherits it. The generated Tools page carries the slot census at the current pin.

Layer Ids Purpose Categories
CORE — system substrate, no tool categories kernel, hardening, safety defaults, desktop(s), CLI/menu engines
CLASSIC C-1…C-10 traditional software security assessment (Kali heritage) Information Gathering · Vulnerability Analysis · Web App Assessment · Database Assessment · Password & Offline-Credential · Reverse Engineering · Signal & SDR (opt-in) · Exploitation & Post-Exploitation · Forensics/IR · Runtime AppSec for AI
ATTACK A-1…A-8 AI offensive tooling, MITRE ATLAS-mapped AI Recon · LLM Prompt Injection & Jailbreaking · Agentic/MCP & Tool-Use · Model Supply Chain · Model Extraction/Inversion · Multi-modal · AI System Exploitation · Data Poisoning
DEFENCE D-1…D-7 AI defensive and assurance tooling Runtime Defence & Guardrails · Detection & Observability · Evals & Benchmarks · Provenance & Signing · Threat Modelling · Governance Mapping · Standards Verification
OPS O-1…O-7 the platform the other layers attack and defend MLOps · Compute & Serving · Vector Stores · Local LLM · SDKs & Dev Env · Docs & Authoring · Agent & MCP Operability

One manifest generates the navigation. Three views mirror the practitioner’s three questions: layer-first (the menu), tactic-first (the MITRE ATLAS matrix), and standard-first (the ASVS / AISVS / MLSVS / OWASP LLM Top 10 evidence views).

Architecture

Two repositories and a website mirror the nixpkgs↔︎NixOS split:

  • kailash-packages — the overlay: every bespoke tool derivation, the tool manifest (tools.yaml + categories.yaml), update automation, own CI. Independently consumable on any NixOS host.
  • kailash-os — the OS: root flake, the category module tree, profiles, images, the kailash CLI, tests, release engineering.
  • kailash-website — this site: generated docs and governance pages, built from a pinned manifest revision.

One manifest drives six system views: overlay, module tree, desktop menu, CLI, docs, and the verification evidence matrix. No two surfaces drift. Every category is an opt-in NixOS module (kailash.layers.<layer>.categories.<id>), generated from the manifest, VM-tested, never hand-maintained.

Releases are lockfile states: the lockfile is the version. Releases tag 2026.01-alpha.1 first (staged cuts); bare 2026.01 stays reserved for GA. The two-digit release index counts releases within the year (01, 02, …, 2026.10, …). Nixpkgs inputs refresh monthly. Tool updates arrive as reviewed PRs by staleness class. A release train departs every 6 weeks with its images, checksums, and signed changelog.

The lab thesis

The platform and the attack tooling ship together. OPS deploys MLflow, vLLM, TorchServe, vector stores, and agent runtimes as documented lab configurations. ATTACK carries the platform-exploitation probes validated against those same deployments. Builders learn to break; breakers learn to build. One pinned environment carries the distribution’s own verification harness.

Verification standards

Assurance work binds to the OWASP stack: ASVS 5.0, AISVS 1.0, MLSVS / ML Top 10, and both epochs of the OWASP LLM Top 10 (2026 and v2-2025). Each standard chapter points to the tools and data packs that test it. kailash verifier records the chapter-by-chapter evidence matrix (KA-35); the claim “verified against the standard” then names a runnable command.

Governance

Kailash ships offensive AI-security tooling for authorised security work only — red-team engagements with written authorisation, academic research, and defender validation of one’s own systems. Users stay responsible for the law in their jurisdiction (in Australia: Cyber Security Act 2024 (Cth); Criminal Code Act 1995 (Cth), Div 474).

Safety is part of the OS rather than documentation beside it. KAILASH_LAB_MODE defaults off. The OS double-gates active-exploit tooling behind explicit host and shell opt-in: the host key kailash.labMode = true plus the shell key KAILASH_LAB_MODE=1 per invocation. No service auto-starts. sops-nix protects the secrets, and the system reads them at runtime.

Roadmap

The first release train ran: 2026.01-alpha.1 published on 13 October 2026. The tag carries the full four-layer catalogue on a pinned NixOS base; the generated Tools page states the census at each pin. Every release artefact carries SLSA provenance and an SBOM attestation. The docs estate is live: the Kailash OS Guide and the Kailash OS Cookbook.

The Roadmap page carries the project timeline and the release-train plan: every item with its tracking issue and evidence. The Kailash Roadmap project board carries the work as issues KA-NN in kailash-os/kailash-os. The dependency graph is the canon map.

Documentation

The docs estate is three sites:

  • kailash.site — project home, paper, and roadmap (this site)
  • docs.kailash.tools — the user and administrator guide, with a man-style reference page for every tool in the catalogue
  • book.kailash.tools — the cookbook: task-oriented recipes in the Problem → Solution → Discussion → See-also format

The guide explains what a tool is; the cookbook shows how to use it.

Paper

The paper “Kailash OS — An Operating System for Building and Securing AI” records the design. It carries the taxonomy derivation from MITRE ATLAS and the OWASP standards, the manifest-driven architecture, the release engineering of a reproducible OS, and the explicit scope exclusions.

Read the current revision: kailash-paper.pdf (Revision 1.5, 5 November 2026 — updated as the implementation progresses; the live copy reflects the latest revision).

The implementation specification carries the phase gates, the packaging shapes, the manifest schema, and the verification bindings. It is docs/spec.md in the project repository.

Get involved

Kailash is open source under BSD-3-Clause. Bundled tools keep their upstream licences. The packaging harness, the taxonomy, and the CLI accept contributors: add a tool via a manifest entry, not via hand-edited module files. CI validates, builds, and tests every change. Sign every commit. Open an issue on the roadmap board, or pick up a 2026.01 train item.

Contact: Shain Singh — shain.singh@owasp.org (OpenPGP: 9CB1781DC2BB28D37A0155DCAAF8226F3F4C1712)