Skip to content

Latest commit

 

History

38 Commits

Folders and files

NameName
Last commit message
Last commit date
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Doldskrift

Doldskrift Sweden

CI License MSRV crates.io docs.rs GitHub Pages site thanks.dev Protocol

Machine-native visual information.
A platform for encoding structured data into a visual writing system built for deterministic machine reconstruction.

Swedish etymology: dold (hidden / concealed) + skrift (writing / script)
→ “hidden writing” / “concealed script” · Maskinskriven betydelse — inte hemlighet.
Machine-written meaning — not secrecy. Hidden ≠ encrypted.

Open and Neural are not encryption. Visual illegibility and learned grammars are not confidentiality. Protected is AEAD + Gate authorization when it ships — glyphs encode ciphertext. Today: refuse stubs only. Until then: encrypt with established cryptography first, then optionally represent the result with Open-mode Doldskrift.


Product

Doldskrift is an independent open-source platform for machine-native visual information: protocol, glyphs, recognition, Lab CLI, demos, and documentation — one architecture with three profiles.

It is designed for agents, tooling, and research workflows that need a stable visual carrier: encode structured information, render it as machine glyphs, transmit or display the artifact, then reconstruct the original bytes with deterministic fidelity (Open) or a declared reader path (Neural).

dold guide           # in-CLI onboarding
dold about           # etymology + profiles
dold museum          # protocol archaeology

Why it exists

Humans and machines do not share the same reading stack. Doldskrift treats the visual plane as a first-class information channel for machines: geometry, protocol framing, and reconstruction — not Latin text decoration, and not a substitute for cryptography.

The official logo is glyph DSK_PROJECT_MARK (U+E1F0) — a real machine glyph, not ornament.


Architecture: Open · Neural · Protected

DOLDSKRIFT
┌──────────────────────────────────────┐
│ Semantic Layer                       │  structured machine information
├──────────────────────────────────────┤
│ Protocol Layer                       │  frames, streams, schemas, sessions
├──────────────────────────────────────┤
│ Visual Layer                         │  glyphs, alphabets, rendering
├──────────────────────────────────────┤
│ Recognition Layer                    │  detection, vision, reconstruction
└──────────────────────────────────────┘
Name Role
DSK Protocol (DSK/1 stable, DSK/2 experimental, DSK/3 Neural + Protected)
LSG Latent Structure Grammar (Neural intermediate)
DSK-R Neural readers (DeterministicBaseline / Mock today)
Gate Authorized runtime for Protected objects
MGE Machine Glyph Engine
DVE Doldskrift Vision Engine
Forge Alphabet / glyph optimization
Lens Flagship visual demonstration
Lab Developer tooling (dold)

Profiles

Profile What it is Shipping status
Open Deterministic codec + glyphs (DSK/1–2). Visual concealment, not secrecy. Yes (prototype)
Neural Learned LSG + compatible reader (DSK/3). Fail closed on unknown grammar. Not encryption. Foundation stubs + DeterministicBaseline
Protected AEAD ciphertext + Gate authorization. Glyphs encode ciphertext. Refuse stubs only — no AEAD yet
DATA → DSK → MGE → VISUAL MEDIUM → DVE → DSK → DATA

Demo: humans vs agents

Humans see abstract glyphs; agents recover plaintext

Reader Experience
Humans Structured abstract marks — geometry, not Latin
Agents Deterministic decode: symbols → bytes → original data

Encode pipeline

Specimen

Input:

Deploy worker seven.

Doldskrift rendering of “Deploy worker seven.”

Decode recovers the exact original string.

Modes:

Visual, encoded, and session modes


Install

From crates.io (recommended)

cargo install doldskrift-cli
dold doctor

Library (Cargo.toml):

[dependencies]
doldskrift = "0.5"

Related crates: doldskrift, doldskrift-font, doldskrift-vision, doldskrift-cli, doldskrift-bindings.

API docs: docs.rs/doldskrift.

From source

git clone https://github.com/theworker02/doldskrift.git
cd doldskrift
cargo install --path crates/doldskrift-cli
dold doctor

MSRV: Rust 1.75+.


Quick start

dold guide                  # product orientation
dold doctor                 # install health + constants sync
dold self-test              # encode/decode/validate + protected refuse
dold demo -o ./dold-demo    # .dsk + postcard + mark pack
dold commands               # curated catalog (alias: topics)
dold version --verbose

echo "Hello from Doldskrift" | dold encode > message.dsk   # alias: dold e
dold pipeline --text "Hello from Doldskrift" -o message.dsk
dold validate message.dsk
dold inspect --json message.dsk
dold decode message.dsk     # alias: dold d

dold convert --from text --to postcard --text "Deploy worker seven." -o card.svg
dold convert --from postcard --to dsk -i card.svg -o card.dsk

dold logo --out assets/brand/logo-mark.svg
dold config init

# Shell completions (optional)
dold completion powershell > _dold.ps1

# Agent-friendly errors
$env:DOLD_JSON_ERRORS = "1"   # PowerShell → stderr doldskrift.error/1

End-to-end paths

  1. Open: encodevalidate.dsk / MGE specimen → Lens or Surfaces → decode
  2. Neural: dold neural encoderender (SVG carrier) → reconstruct (DeterministicBaseline)
  3. Protected: encode --mode protectedinspect shows Readable: No → decode / open refuse (AEAD not shipping)

Interactive demos — static GitHub Pages (no separate hosted web app):

Live site: https://theworker02.github.io/doldskrift/

npm install
npm run site:preview   # build site/dist + local static preview

Lab CLI highlights

The dold binary is the Lab surface for developers and agents.

Area Commands
Onboarding guide, about, museum, demo, commands / topics
Codec encode / e, decode / d, pipeline, validate, fmt, inspect, convert
Health doctor, self-test, version --verbose, schema, completion
Surfaces postcard, ambient, seal, radio, mesh, kaleidoscope, timeline, …
Neural neural, dataset, reader, evaluate
Brand / funding logo, brand, funding

Unexpected Open surfaces still obey the honesty model (Open ≠ encryption; hashes ≠ signatures). Gallery: Surfaces. Full list: dold surfaces list / dold guide surfaces.

Command What it does
dold (no args) Tiny braille / project-mark fingerprint
dold radio exchange "ping" Two-agent visual packet
dold echo / --verify Self-describing document + SHA-256 glyph strip
dold ambient / --decode Constellation SVG Open channel
dold postcard / --read Machine postcard
dold convert Bridge text/dsk ↔ postcard/ambient/seal
dold render --tty Braille / block art without a browser

Libraries

Rust (published on crates.io as 0.5.0):

[dependencies]
doldskrift = "0.5"
use doldskrift::{encode, decode, Mode};

let opaque = encode("Hello, agent!").unwrap();
assert_eq!(decode(&opaque).unwrap(), "Hello, agent!");
let _ = Mode::Encoded;

JavaScript — primary SDK @doldskrift/core; thin @doldskrift/web / @doldskrift/react adapters; generated WASM (doldskrift-bindings). Rust owns the protocol truth. npm packages are versioned in-repo and are not the crates.io release channel.


Documentation & links

Resource Link
GitHub repository theworker02/doldskrift
GitHub Releases Releases
GitHub Pages theworker02.github.io/doldskrift
Docs portal docs/
CLI reference docs/reference/cli.md
Architecture ARCHITECTURE.md
Roadmap ROADMAP.md
Security SECURITY.md
Brand kit docs/brand.md
Changelog CHANGELOG.md
crates.io doldskrift
docs.rs doldskrift
Funding thanks.dev/u/gh/theworker02

Lens / Surfaces / Neural demos: Lens · Surfaces · Neural


Repository layout

crates/          # Rust workspace (core, cli, vision, font, bindings)
packages/        # @doldskrift/core (+ thin web/react) · WASM bindings
site/            # GitHub Pages demos
docs/            # Product + protocol + neural docs map
spec/            # Normative DSK/1 + protocol-constants.json
training/        # Python experiments only (not a second protocol)

Status (honest)

Component Maturity
DSK/1 codec + .dsk Stable prototype
DSK/2 semantics Experimental
DSK/3 Neural (LSG + readers) Foundation stubs (ADR-0007)
DSK/3 Protected + Gate Design + refuse stubs (ADR-0006)
MGE glyph engine Candidate (evolving)
DVE vision Experimental
WASM / CLI / Lens Beta / shipping on Pages

Open-mode visual illegibility is not a security boundary. See SECURITY.md.


Support / Funding

Bugs and documentation issues: SUPPORT.md. Commercial collaboration: Partners.


Contributors

Curated project contributors (not GitHub's automatic commit graph):

Name Role
theworker02 Creator and maintainer

Bots such as Dependabot are not listed as project contributors. Full note: CONTRIBUTORS.md.


Contributing

See CONTRIBUTING.md and CODE_OF_CONDUCT.md.

Issues and pull requests are welcome for protocol clarity, Lab DX, documentation honesty, and Open-mode reliability. Please keep Neural and Protected claims aligned with what actually ships.


License

Licensed under either of:

at your option. See LICENSE.

About

Encode structured information into a visual writing system designed for deterministic machine reconstruction.

Topics

Resources

Code of conduct

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Sponsor this project

Packages

Contributors

Languages