LuSaur Systems / AI governance & security

Control access.
Trace decisions.
Govern AI.

AI introduces new questions about information access, evidence, and responsibility. Our public technology projects help teams define controls and build knowledge workflows that can be reviewed.

Governance & security

Start with the
operating requirement.

Who can retrieve a record? Which source supports an answer? What needs human review before an agent acts? These questions guide the specification and software described below.

Choose by the work you need to do

Client / Use case / Product

Suggested starting points based on the documented scope of each repository. These are audience recommendations, not a list of existing clients.

Client / Use case / Product
ClientUse caseProduct
AI governance leads, risk teams, and enterprise architectsDefine evidence, review, and authorization requirements for AI-assisted decisions.GKOS Standard
Platform engineers, security architects, and AI application developersValidate knowledge, trace citations, and restrict what local applications and agents can retrieve.GKOS Engine
Independent developers, technical researchers, and small knowledge teamsCheck, index, search, and export Markdown folders from scripts without Obsidian.GKOS Engine Lite
Researchers, analysts, and teams working with connected notesExplore relationships and retained history in a Markdown knowledge collection.Kosmos-Oden
Individual Obsidian users, writers, students, and researchersNavigate a personal vault and add structured metadata through a previewed formatting workflow.Kosmos Research Studio Lite

Public portfolio / Reviewed September 5, 2026

Five projects. Defined roles.

Conceptual architecture
Public pre-standard

GKOS Standard

A governance specification for tracing how information becomes a decision or an action. GKOS defines records for sources, claims, checks, review, context, and authorized use. It gives governance and engineering teams a common structure for specifying what an AI workflow must retain and when it must stop.

Developmental Product details
Conceptual architecture
Core tooling; service contracts integration-only

GKOS Engine

A TypeScript engine, command-line tool, and local service for governed Markdown knowledge. It validates records, resolves lineage, builds graphs, and supports indexed search with citations checked against source files. Its local service applies credential-based access to knowledge views and exposes a bounded set of read-only MCP tools.

Source implementation Product details
Conceptual architecture
CLI; desktop qualification in progress

GKOS Engine Lite

A focused command-line interface that delegates validation, indexing, assessment, search, graph creation, and export to a pinned GKOS Engine. It targets everyday note collections and human-editable metadata. Optional AI assistance produces candidate suggestions for review; it does not edit notes.

Source tooling Product details
Conceptual architecture
Standalone viewer and Obsidian plugin

Kosmos-Oden

A visual knowledge application that presents folders and linked notes as an explorable astronomical map. It offers search, focused navigation, and time-based views. The standalone viewer can open a local folder offline; the Obsidian edition adds vault integration and knowledge tools.

Repository product line Product details
Conceptual architecture
Conservatively maintained Obsidian edition

Kosmos Research Studio Lite

The maintained Frozen Core edition of the Vault Kosmos line combines a 3D note viewer with readable GKX metadata. Formatting uses previewed changes and backups. It also documents a local read-only Agent API, with sensitivity settings controlling which notes can be exposed.

Frozen Core Edition Product details

Security priorities

Controls where information is used.

01 /

Access

Restrict returned information to the identity, purpose, and permissions of the request.

02 /

Provenance

Retain source relationships and check citations against the material they reference.

03 /

Review

Keep model suggestions distinct from approved decisions and actions.

04 /

Local operation

Evaluate knowledge tooling locally and enable optional AI or service interfaces deliberately.

Evaluate the specific product and version.

The standard defines responsibilities; the engines implement bounded capabilities; Kosmos provides visual knowledge tools. The linked repositories document release status and integration limits. Product descriptions do not establish certification or compliance.

What we are working toward

Future Horizon.

Explore Future Horizon
Conceptual architecture
Agent infrastructure

AgenticOS

An operating foundation for agents, persistent state, and controlled execution. Marshal is the alpha development track.

Alpha / Development direction Marshal overview
Conceptual architecture
Research workflows

NextGen Governed Research Suite

A future workspace connecting source evidence, AI-assisted research, review, and reproducible records.

Future direction
Conceptual architecture
Local knowledge

Governed Desktop Workspace

Simpler desktop access to local knowledge tools, with explicit permissions and review controls.

Development direction / Qualification ahead

Work with LuSaur

Discuss your project.

Tell us where AI enters your workflow, what information it uses, and which decisions need oversight.

Contact LuSaur