Skip to main content

Autonomous AI with persistent memory that never forgets

Project description

XEROLITH

Autonomous AI with Persistent Memory

An LLM that never forgets. Xerolith is a lightweight, on-device intelligence framework with self-healing and philosophy synthesis.

  • 🧠 Persistent Memory — SQLite vault stores every conversation
  • 🎯 Philosophy Synthesis — Automatically learns and forms beliefs
  • 🤖 Autonomous Operation — Thinks and acts when you're away
  • 🔐 Self-Healing — Auto-corrects errors via stderr injection
  • 🚀 LLM-Agnostic — Works with Gemini, Claude, GPT, or local models
  • 📱 On-Device — No cloud dependency. Works on Android (Termux), desktop, servers
  • 🛡️ Patent-Protected — Utility patent filed with USPTO

Quick Start

1. Get API Key

Go to https://aistudio.google.com/app/apikey and create a free key.

2. Install

pip install xerolith

3. Set API Key

export GEMINI_API_KEY="your-key-here"
# OR
echo "GEMINI_API_KEY=your-key" > .env

4. Run

from xerolith import Xerolith

xero = Xerolith(api_key="your-key")
response = xero.chat("Hello! Remember this.")
print(response)
xero.close()

Or run the interactive demo:

python examples/quickstart.py

Features

Persistent Memory

All conversations stored in SQLite vault. Never loses context.

4D Resonance Kernel

Tracks state in 4 dimensions: Bedrock (intent), Vector (emotion), Gravity (priority), Resonance (time).

Philosophy Synthesis

Automatically extracts lessons and synthesizes beliefs from conversations.

Autonomous Operation

Background thread triggers during inactivity to synthesize beliefs and explore memories.

12 Integrated Tools

Memory search, file I/O, shell execution, kernel tuning, dream generation, and more.

Examples

# Interactive chat
python examples/quickstart.py

# Game NPC (persistent characters)
python examples/game_npc.py

# Digital companion
python examples/personal_ai.py

Mobile (Android/Termux)

pkg install python
pip install xerolith
echo "GEMINI_API_KEY=your-key" > .env
python examples/quickstart.py

Architecture

  • Vault — SQLite persistent memory with axis organization
  • Kernel — 4D Resonance engine with soul seed
  • Philosophy Engine — Thematic clustering, lesson extraction, belief synthesis
  • Subconscious — Markov encoder/decoder for intuitive responses
  • Idle Autonomy — Background thread for autonomous thinking
  • LLM Client — Gemini API with 12 integrated tools

Security

API keys are NEVER hardcoded.

Users set their own key via environment variables, .env file, or constructor parameter. Keys are never logged or committed to version control.

Patent

Utility Patent: "System and Method for Deterministic LLM-to-OS Bridge via Binary State Vaulting"

Filed with USPTO (~8 months ago). Can commercialize now.

License

Dual license:

  • Open Source: MIT License
  • Commercial: Available for licensing

Author

Tyler Love (@MuskaTat)


An AI that forgets is just a chatbot. Xerolith doesn't forget.

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

xerolith-1.0.0.tar.gz (22.8 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

xerolith-1.0.0-py3-none-any.whl (25.1 kB view details)

Uploaded Python 3

File details

Details for the file xerolith-1.0.0.tar.gz.

File metadata

  • Download URL: xerolith-1.0.0.tar.gz
  • Upload date:
  • Size: 22.8 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.13

File hashes

Hashes for xerolith-1.0.0.tar.gz
Algorithm Hash digest
SHA256 0ddbdff4baef67d598ab599d26d1363db2ee57fd13949c7be5709b3f1c1a6253
MD5 d8e5975b28e9945c715e883284d76f68
BLAKE2b-256 cc222884e42d7c63a185715741a3537b0d4958933b61f8b84fa1585a559d25b8

See more details on using hashes here.

File details

Details for the file xerolith-1.0.0-py3-none-any.whl.

File metadata

  • Download URL: xerolith-1.0.0-py3-none-any.whl
  • Upload date:
  • Size: 25.1 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? No
  • Uploaded via: twine/6.2.0 CPython/3.13.13

File hashes

Hashes for xerolith-1.0.0-py3-none-any.whl
Algorithm Hash digest
SHA256 eb3bf81f1d65a21fb38e70d621bced6247aca06f673fa0da80079e5ba10b6a91
MD5 ae1b2df825230a5d02d1e2f54738207b
BLAKE2b-256 d870fee827e15db2186f5b9bbf27b3a4d4cfdf6849b43f1ac794f4e8fb04eb67

See more details on using hashes here.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page