Safe archive workspaces

ITHZ Drive turns archives into verifiable workspaces.

A compact product direction for safe editable archives, native tooling, deterministic agent memory, and carefully published research evidence.

.ithz One container for files, metadata and verification.
Drive Open an archive like a practical workspace.
MCP Deterministic memory packs for AI agents.
Native C++ engine, CLI, safe verify and Windows workflows.

Drive first, format underneath.

ITHZ is shaped as an ecosystem: a product people can try, an engine developers can inspect, and a research path that stays separate from commercial promises.

Windows first

ITHZ Drive

A safe editable archive drive for Windows. Built around inspection, safe extraction, hash preconditions and a clear rollback-minded workflow.

Explorer workflow Editable mount Safe verify
Open section page
Engine layer

ITHZ Native

The C++ CLI and runtime layer: pack, inspect, list, verify, extract and update archives with deterministic gates and reproducible release builds.

C++20 AVX2/scalar CLI/SDK path
Open section page
Agent memory

ITHZ MCP

Deterministic project memory for AI agents: workflow profiles, decisions, gates, risks, context packs and task checkpoints stored in a verifiable archive zone.

Context packs Audit trail Memory zones
Open section page
Research track

Experiment

Kernel and codec experiments are published carefully: reproducible diagnostics, evidence logs and research notes, not premature proof or inflated benchmark claims.

Kernel notes Phase gates Cautious claims
Open section page
Free preview

Download the Windows alpha build.

For personal and non-commercial evaluation. Advanced editions will be added gradually.

v0.1-alpha-rc1 ithz-drive-preview-v0.1-alpha-rc1-win64.zip · 1.4 MB Download ZIP
Public story

What can be said now.

Safe archive UX is the product.

ITHZ Drive can be presented as an editable, verifiable archive workspace for Windows projects and documents.

Native is the engineering base.

The native engine can be shown as the CLI and runtime layer behind inspect, verify, extract, update and packaging workflows.

MCP is the AI memory angle.

Agents can request compact, deterministic context packs instead of repeatedly scanning entire projects or long histories.

Experiments stay honest.

The kernel line can publish dated evidence, negative controls and current candidate mechanisms without claiming final physical proof.

The research direction is simple to state and hard to prove: find structure that survives the order in which a process is read.

Built around verification, not hype.

ITHZ is designed as a product, engine, memory layer and research archive with clear boundaries between what is ready to use and what is still experimental.

Publish

Product names, workflow diagrams, install notes, screenshots, demo archives, dated research notes and reproducible diagnostic summaries.

Hold back

Premature proof language, exact unpublished patentable mechanisms, private datasets, credentials, and unsupported compression benchmark claims.