AI Autonomic Synthesis Platform (v33.0)
A fully autonomous AI CLI platform for task management, trading, project tracking, and automated workflows.
This script is designed to run from ~/bin/ai and integrates multiple AI models, persistent caching, hashing, and streaming output.
Table of Contents
-
Features
-
Installation
-
Directory Structure
-
AGI Models & Loops
-
Tools
-
Commands
-
Hashing & MIME Handling
-
Databases
-
Security
-
Logging & Feedback
Features
AGI Workflow with 12 loops
Streaming mode always enabled
Models: core, loop, 2244-1
Hashing: strings, files, directories, salted
MIME detection and file processing
Project ingestion & rehashing
QBit task management
Bitcoin market analysis & trading
WebKit repository build
Persistent caching & memory
Full logging and error tracking
Installation
- Clone or copy the ai.sh script to ~/bin/ai:
mkdir -p ~/bin cp ai.sh ~/bin/ai chmod +x ~/bin/ai
- Ensure your path includes ~/bin:
export PATH="$HOME/bin:$PATH"
- Run ai --help to verify installation:
ai --help
Directory Structure
All files and databases are stored under ~/.local_ai:
.local_ai/ ├── logs/ # System logs ├── tmp/ # Temporary files ├── swap/ # Compressed large outputs ├── agent_core.db # AI core memory & tool logs ├── ai_task_manager.db # Task manager DB (projects, file hashes, events) ├── secret.key # HMAC secret key
Projects ingested via ai ingest are tracked here.
AGI Models & Loops
Models used:
core → Messenger / state analysis
loop → Planner
2244-1 → Executor / decision-making
AGI loops: 12 maximum per workflow
Streaming output: tokens appear live in terminal
Caching: semantic prompt hashing with persistent storage
Tools
Project Management
Ingest repositories
Rehash and verify files
MIME-based processing
QBit
Log structured tasks to SQLite
BTC Trading
Analyze market (RSI, MACD, support/resistance, volume)
Buy/sell placeholders
WebKit Builder
Clone & build WebKit locally
Logs build output
Market Analysis
Technical crypto analysis for any asset
Commands
ai Run AGI workflow ai ingest Ingest project for tracking ai rehash Rehash project files ai task Create AI task ai qbit QBit task ai btc [action] BTC trade/analyze ai webkit Build WebKit ai analyze Market analysis ai --help Show this help
Fallback: Unknown commands automatically trigger AGI workflow.
Hashing & MIME Handling
String hashing: SHA256, salted options
File hashing: SHA256, SHA512, MD5
Directory hashing: recursive SHA256 of all files
MIME type detection:
text/* → hash & track
image/* → log for AI processing
application/zip → archive logged
Unknown types trigger warnings
Databases
Task DB (ai_task_manager.db):
projects — project hash, path, timestamp
file_hashes — file integrity tracking
events — AI tasks, QBit, other events
Core DB (agent_core.db):
memories — cached AI responses
tool_logs — execution logs of all tools
Security
HMAC secret key stored in secret.key
SQLite & log files can be secured via chmod
Data integrity verified via HMAC calculations
Logging & Feedback
Color-coded terminal output: INFO, WARN, ERROR, SUCCESS
Logs both to terminal and logs/system.log
AGI workflow loops and tool actions fully logged
Supports streaming live output for AI decisions
Example Usage
ai "Create a React component for dashboard"
ai ingest ~/projects/myapp
ai rehash <project_hash>
ai qbit "Generate test plan"
ai btc analyze
ai webkit
ai analyze ETH/USD
This README.md now documents all capabilities, commands, paths, models, tools, and security measures in a clear and structured format for developers or operators.