AngryDroid — Cyberpunk Capsule Hub

angrydroid.neocities.org

86,273 views
223 followers
931 updates
0 tips
angrydroid updated their site.
7 months ago
3 likes
angrydroid's avatar angrydroid 7 months ago

MYTHIC AI DUAL MIND · NEURAL 4:14:31 PM ✨ MYTHIC AI · DUAL MIND active. Neural network ready. 🧠 Primary Mind online. Toggle Second Mind for collaboration. 🔵 Loading Primary Mind (Flan-T5)... ✅ Primary Mind online. 🟠 Loading Second Mind ✅ Second Mind online. Toggle Dual Mind to collaborate. 📀 messages loaded 🧠 neural pathways ready

1984 BBC Micro Model B with a fully functional graphical OS inside its green phosphor CRT. The interface includes a beige case, a curved monitor bezel, scanline overlays, and a menu bar with clickable app icons that load external retro‑themed websites directly into the screen—all styled to look like they are running natively on the vintage machine.
1 like
angrydroid updated their site.
7 months ago
3 likes
Prompt Monkey: This tool is ideal for prompt experimentation, adversarial testing, and teaching prompt engineering concepts in a self‑contained, shareable format.
angrydroid updated their site.
7 months ago
1 like
angrydroid's avatar angrydroid 7 months ago

Local 77M Transformer model for chat and intent classification. Learned facts stored in localStorage (👍 to save, 👎 to correct). Conversation memory (last 30 exchanges). Auto‑learning web search – every web answer is saved for offline recall.

Mythic AI Simulation is Future-terminal in Mythic os enjoy :)
1 like
Auto‑Learning Feedback Loop After every AI answer, 👍 and 👎 buttons appear. Thumbs Up: Stores the question–answer pair as a permanent fact in localStorage. Future answers will reference learned facts. Thumbs Down: Opens an input field to correct the answer. The corrected fact replaces the old one. All learned facts persist across page reloads and are injected into the generative AI’s prompt for b
1 like
📦 Technical Stack HTML5, CSS3, JavaScript (ES Modules) Transformers.js for on‑device LLM inference LocalStorage for memory persistence Fetch API with public CORS proxies and open APIs Web Crypto API for SHA‑256 file hashing The result is a self‑contained, privacy‑friendly AI assistant that remembers conversations, searches the live web, fetches code, and performs local security analysis—all
2 likes
The local AI model requires ~200 MB of RAM and may take 10–15 seconds to load on first visit.
2 likes
Learned patterns (via the transformer model) Generalization (handling unseen inputs) Generative behavior (novel responses) Internet‑connected knowledge (DuckDuckGo & Wikipedia) Code retrieval (Stack Overflow) All within a single, portable HTML file that requires no backend, no API keys, and no cost to deploy. It's a fully functional, memory‑enhanced AI assistant that fits the "Mythic AI"
2 likes

Website Stats

Last updated 5 months ago
CreatedOct 22, 2025

Tags

os tools technology encrypted programing