KameleonDB Skill

Agent-native database for dynamic schema evolution

This OpenClaw skill provides agents full control over data lifecycle: schema design, ingestion, evolution, and querying - all without migrations or DDL.

Features

• Dynamic Schema Evolution - Add fields at runtime, zero lock operations
• Agent Hints Pattern - Queries return optimization suggestions inline
• Hybrid Storage - Flexible JSONB or dedicated typed tables
• Full Audit Trail - Every schema change tracked with reasoning
• JSON I/O - All operations support --json for machine-readable output

Installation

pip install kameleondb[postgresql]

Configuration

Set KAMELEONDB_URL in OpenClaw config or environment:

{
  "env": {
    "KAMELEONDB_URL": "postgresql://localhost/kameleondb"
  }
}

For development, use SQLite:

{
  "env": {
    "KAMELEONDB_URL": "sqlite:///./kameleondb.db"
  }
}

Quick Start

# Initialize
kameleondb admin init

# Create entity
kameleondb --json schema create Contact \
  --field "name:string:required" \
  --field "email:string:unique"

# Insert data
kameleondb --json data insert Contact '{"name":"Alice","email":"alice@example.com"}'

# Query
kameleondb --json data list Contact

See SKILL.md for complete documentation.

Use Cases

• Data Ingestion Agents - Discover schema needs, create entities, batch insert
• Enrichment Agents - Add fields as new data sources discovered
• Query Agents - Generate SQL with schema context, optimize based on hints
• Data Modeling Agents - Design relationships, materialize for performance

Why KameleonDB?

Traditional databases force agents to work within rigid schemas designed by humans. KameleonDB makes agents the data engineers - they design schemas, evolve them as they reason, and optimize storage based on usage patterns.

Schema-on-Reason: Schema emerges from continuous agent reasoning, not upfront human design.

Links

• GitHub Repository
• PyPI Package
• First Principles
• Architecture Guide