Bonnet CLI

A Python CLI tool for managing structured knowledge base (memory) and preparing highly compressed XML context using a knowledge graph architecture.

Installation

pip install -e .

Usage

The CLI can be run in two ways:

• bonnet (if installed and in PATH)
• python -m bonnet (from the project directory)

Dependencies

• Python 3.7+
• Click 8.0.0+
• python-dateutil 2.8.0+
• pydantic 2.0.0+

Usage Examples

Store a topic (master ENTITY record)

# With explicit ID
bonnet topic --id M1 "The Meaning of Life"

# With auto-generated simple ID
bonnet topic "The Meaning of Life"  # Creates ID: T1
bonnet topic "Machine Learning"     # Creates ID: T2

Store attributes

# Store a fact
bonnet attr --about M1 --type FACT --subject "Answer" "42"

# Store a reference
bonnet attr --about M1 --type REF --subject "Source" "The hitchhikers guide to the galaxy"

Search and generate context

# Search for entities and generate context
bonnet context --about M1

# Search by text (will prompt for disambiguation if multiple matches)
bonnet context --about "life"

Output Format

The context command generates XML output in the following format:

<context>
  <entity id="M1">
    <name>The Meaning of Life</name>
    <relationship type="has_attribute" to="A1">
      <content>FACT:Answer:42</content>
    </relationship>
    <relationship type="has_attribute" to="A2">
      <content>REF:Source:The hitchhikers guide to the galaxy</content>
    </relationship>
  </entity>
</context>

Architecture

Bonnet uses a knowledge graph architecture with:

• Entities: Master records representing topics or concepts
• Attributes: Facts, references, tasks, or rules linked to entities
• Nodes: Graph nodes representing both entities and attributes
• Edges: Relationships between nodes in the knowledge graph
• FTS Search: Full-text search across the knowledge base using SQLite FTS5