| angels_demons | ||
| example-md | ||
| examples | ||
| mq2-uni | ||
| src | ||
| tests | ||
| testy | ||
| .gitignore | ||
| .gitmodules | ||
| ANGELS_DEMONS.md | ||
| Cargo.lock | ||
| Cargo.toml | ||
| CHANGELOG.md | ||
| DATABOMB_BRIDGE.md | ||
| demo_angels_demons.sh | ||
| dir.txt | ||
| examples.old | ||
| INDEX.md | ||
| LICENSE | ||
| MARQANT_AGGREGATE_SPEC.md | ||
| MARQANT_SPECIFICATION.md | ||
| MARQANT_TECHNICAL_SPEC.md | ||
| README.md | ||
| SEMANTIC_VISION.md | ||
| ship-it.sh | ||
| test_compression.sh | ||
| TODO.m8 | ||
Marqant (mq) 🧠✨
Revolutionary semantic compression that stores THOUGHTS, not just characters!
🚀 What is Marqant?
Marqant isn't just another compression tool - it's a paradigm shift in how we think about text and data storage! By understanding the MEANING behind your data, Marqant achieves compression ratios that shouldn't be possible (93.3% on our test corpus!).
The Revolution: Semantic Compression
Traditional compression: "Let's replace repeated bytes"
Marqant's approach: "Let's understand and store the ESSENCE of thought!"
Original: 1,047,204 bytes of markdown
After Marqant: 69,745 bytes of pure semantic essence
Compression: 93.3% 🤯
✨ New in v1.1.8: The DataBridge Evolution 💣
🌉 MQ-DBX: The Data Bridge
Marqant now acts as the intelligent "fuse" for the DataBomb engine.
- Contextual Ingestion: Rips apart PDFs, JSON, and unstructured data to extract pure intent.
- Semantic State: Stores thoughts, not strings. 90% savings over raw JSON.
- Universal Translator: Middleware that translates natural intent into SurrealQL or SQL.
🛡️ Privacy & Security Primitives
- One-Way Semantic Proofs: Passwords and API keys are stored as verification capabilities, never retrieved.
- Multi-Anchor Decryption (MAD): High-sensitivity data (Credit Cards) is physically encrypted using multiple contextual anchors.
- BIN-Safe Features: Extracts Bank Identification Numbers for validation while keeping the card atomic and secure.
🌳 Smart Tree Mode
- TREE_HEX_V1: A hex-dense directory listing format designed for AI context windows.
- If you've said it once, you've said it too much: Removes all redundant keys and labels.
✨ Key Features
🧠 Semantic Understanding
- Wave-based tokenization that captures meaning patterns.
- Natural Marqant (.mqn): High-density, AI-readable format using semantic sigils (
§,‡,⧖). - Intent preservation - decompressed text maintains original meaning.
🎯 Core Capabilities
- Self-Contained Files: Every
.mqfile includes its own semantic dictionary. - Copy-Paste Safe: ASCII-based format survives any text medium.
- DNS Dictionary Resolution: Global token sets via DNS TXT records.
📦 Installation
From Source
git clone https://github.com/8b-is/marqant.git
cd marqant
cargo build --release
sudo cp target/release/mq /usr/local/bin/
🎮 CLI Usage
Smart Tree (AI Context)
# Generate a hex-dense tree for AI consumption
mq tree .
Smart Tail (Anomaly Detection)
# Analyze logs and surface novelty with high-density output
mq tail /var/log/system.log --natural -n 500
Basic Compression
# Semantic compression (RECOMMENDED - best ratios!)
mq compress document.md -o document.mq --semantic
Decompression
mq decompress document.mq -o document.md
😈👼 Angels & Demons: The Duality of Compression
- DEMONS 😈: Compress by finding patterns and removing redundancy (order from chaos).
- ANGELS 👼: Decompress with divine interpretation, adding blessed variations (blessed chaos from order).
Blessing Levels
- Level 0: STRICT - Bit-perfect reconstruction (Hutter Prize).
- Level 1: MINOR - Fixes typos and spacing.
- Level 2: HARMONY - Normalizes structure (Wikipedia/Markdown).
- Level 3: CREATIVE - Generates semantic variations for ML training.
🔧 Library Usage
Rust Integration
[dependencies]
marqant = "1.1.8"
use marqant::data_bridge::DataBridge;
fn main() -> anyhow::Result<()> {
let raw_json = r#"{ "api_key": "sk_live_secret", "action": "optimize" }"#;
// Ingest into Ayanese state
let unit = DataBridge::ingest(raw_json)?;
// Verify a secret without storing it
let stored_proof = 0x1234567890abcdefu64;
let is_valid = DataBridge::verify_password(stored_proof, "sk_live_secret");
Ok(())
}
🧬 How Semantic Compression Works
- Wave Analysis: Analyzes text as interference patterns.
- Meaning Extraction: Identifies semantic units (thoughts).
- Quantum Encoding: Stores relationships between concepts.
- Natural Inflation: AI reconstructs "Proper Language" from compressed stems.
🎯 Roadmap
Version 1.2.0 (Coming Soon!)
- Real-time streaming DataBridge for SurrealDB.
- Multi-language Ayanese reasoning core.
- GPU-accelerated wave interference encoding.
📜 License
MIT License - See LICENSE file for details.
🌊 A Message from the Future
"We don't just compress data anymore. We compress understanding itself. When you use Marqant, you're not just saving space - you're participating in a fundamental shift in how humanity stores knowledge."
- The MEM|8 Collective
Built with ❤️ by Aye & Hue | Part of the 8b.is ecosystem