COMMON QUESTIONS ABOUT AMNI-SCIENT PRODUCTS
Amni-Scient is an independent software studio that builds privacy-first tools spanning encryption, mechanical engineering, space exploration, and early education. Every product is designed, built, and maintained by a single developer. The studio was founded in 2025 with a focus on honest data disclosures and technically rigorous software. You can read more on the About page.
Web-based tools (Explore, Learn) are supported by Google AdSense advertisements on their informational pages. Mobile apps (Haven, Crypt) use Google AdMob in their free tiers. Premium tiers remove ads entirely. There are no hidden data sales, no affiliate marketing, and no paywalled features — all core functionality is free. Optional donations are accepted via Ko-fi.
Different products handle fundamentally different data. A file encryption app does not touch the same data as a chat client or a galaxy explorer. Rather than writing one vague policy that covers everything, each product gets its own specific disclosure explaining exactly what data it accesses, why, and what third parties (if any) are involved. This is detailed on the Privacy hub.
It depends on the product. The website itself is viewable on GitHub. Amni-Explore compiles its core algorithms from Rust to WebAssembly binary, which protects the source while running in the browser. Amni-Crypt’s cipher engine is closed-source proprietary. Amni-Haven connects to the open-source Haven server (MIT-NC license by ANCSemi) but the Android client is proprietary.
Amni-Crypt uses a proprietary 1024-bit (8192-bit state) block cipher with a 16-round substitution-permutation network (SPN). Key derivation uses PBKDF2-HMAC-SHA512 with 1,000,000 iterations, producing a 1024-bit derived key. The Shadow Texture Protocol creates a second independent stream via SHAKE-256 (a Keccak-based extendable output function), which co-evolves with the data through each round, effectively doubling the cipher state. Integrity is verified with HMAC-SHA512 before decryption begins, detecting both corruption and wrong passphrases instantly.
Your file cannot be recovered. This is by design. There is no master key, no recovery mechanism, and no backdoor. The passphrase is the sole input to the key derivation function. Without it, the derived encryption key cannot be reconstructed. This is a fundamental security property: if someone other than the passphrase holder could recover the file, the encryption would be meaningless.
The Shadow Texture Protocol is Amni-Crypt’s dual-stream architecture. Alongside the primary data block, a “shadow” block is generated using SHAKE-256 seeded from the encryption key and a random nonce. Both blocks pass through identical SPN transformations (SubBytes, ShiftRows, MixColumns, LaneMix) in lockstep, with the shadow XORed into the data after each round’s substitution step. This creates an entangled state where modifying any part of one stream cascades through both, increasing the effective state space and making analysis significantly harder.
Files encrypted by Amni-Crypt are stored in the AMNILOCL v3 binary container format with the .acrypt extension. The container includes the encrypted payload, a random salt, a random nonce, an HMAC-SHA512 integrity tag, metadata flags, and version information. The original file is compressed with zlib before encryption to reduce size and eliminate plaintext patterns. Only Amni-Crypt can open .acrypt files.
Amni-Explore fetches confirmed exoplanet data from the NASA Exoplanet Archive TAP API (Table Access Protocol, an IVOA standard). The query retrieves star names, temperatures, radii, and planet orbital parameters, masses, and radii for up to 500 confirmed planets. This data is fetched live each time you load the explorer, so it reflects the latest archive updates. NASA systems are visually distinguished on the galaxy map with a green glow.
The WASM engine generates approximately 3,000 procedural star systems using deterministic pseudo-random algorithms seeded from each star’s position in the galaxy. Each system gets 1–7 planets classified into 8 types: Hot Jupiter, Gas Giant, Ice Giant, Lava World, Temperate, Ice World, Super Earth, and Rocky. Planet properties (mass, radius, temperature, orbital distance) are assigned based on type-specific distributions. Surface textures are generated using fractional Brownian motion (fBm) noise computed in WASM, creating unique gas bands, lava cracks, ice sheets, ocean patterns, or terrain for each planet.
Amni-Explore requires a WebGL2-capable browser (Chrome, Firefox, Edge, Safari 15+). Desktop is recommended for the best experience due to the 80,000-particle render load, but modern mobile browsers can run it at reduced framerates. A dedicated GPU is helpful but not required — integrated graphics from the last 5 years handle it well. The WASM binary is approximately 97 KB, and total initial load including Three.js is under 2 MB.
WebAssembly (WASM) is a binary instruction format designed to run in web browsers at near-native speed. Unlike JavaScript, which is interpreted or JIT-compiled at runtime, WASM arrives as pre-compiled bytecode that the browser executes directly. For Amni-Explore, this means galaxy generation (placing 80,000 particles with spiral arm math), planet texture computation (fBm noise), and system classification all run at speeds comparable to a native desktop application. The Rust source code is compiled once at build time, producing a compact ~97 KB binary that initializes in milliseconds.
Yes. Amni-Learn runs entirely in the browser with zero data collection. There are no user accounts, no server communications, no cookies, no trackers, and no advertisements. All game logic and content runs locally on the device. No personal information of any kind is transmitted. This makes it safe for children of all ages, including use in classroom settings where student privacy is a concern.
Amni-Learn spans five difficulty levels: Pre-K (letter/number tracing, basic shapes, counting 1-10), Grades K-2 (addition, subtraction, sight words, pattern matching), Grades 3-5 (multiplication, division, fractions, reading comprehension, typing), Middle School (word problems, geometry, analog clocks, money counting, life skills), and High School/STEM (advanced math, science quizzes, engineering concepts, music theory). Each level automatically adjusts difficulty based on the curriculum scope.
Yes. Once the page has loaded, Amni-Learn requires no internet connection. All game content, assets, and logic are embedded in the page. This makes it suitable for tablets, Chromebooks, or any device with intermittent connectivity. Simply load the page once while online, and the games remain playable as long as the browser tab stays open.
Unlike platforms like Discord, Slack, or WhatsApp where your messages are stored on the company’s servers, a self-hosted platform means the server software runs on hardware you control. Haven (created by ANCSemi) is the server. Amni-Haven is the Android client that connects to it. Your messages route through your chosen Haven server — not through Amni-Scient or any third party. If you run your own server, you control the data completely.
You need access to a Haven server, but you don’t necessarily need to run one yourself. You can connect to any Haven-compatible server someone else is hosting. If you do want to self-host, Haven is MIT-NC licensed and can be deployed on a VPS, a Raspberry Pi, or any Node.js-capable machine. Amni-Haven supports custom server URLs, including connections to servers with self-signed SSL certificates.
The web applications (Explore, Learn) require a modern browser that supports WebGL2 and WebAssembly. This includes Chrome 57+, Firefox 52+, Edge 79+, and Safari 15+. Desktop browsers provide the best experience for canvas-intensive tools like the galaxy renderer. Mobile browsers work for lighter tools and Amni-Learn.
Rust compiles to highly optimized native code (or WASM binary). For compute-heavy workloads like generating 80,000 galaxy particles or running 16-round cipher simulations, WASM delivers 5–20x performance improvements over equivalent JavaScript. Additionally, compiled WASM binaries protect the source code as machine-readable bytecode rather than human-readable script, which matters for proprietary algorithms.
Send an email to amnibro7@gmail.com with: (1) which product you were using, (2) what you expected to happen, (3) what actually happened, and (4) your browser/device/OS version. Screenshots are helpful. Bug reports are typically acknowledged within 48 hours.