Cobra Agent System

Overview

This agent framework simulates the core operational logic for an advanced, serpentine robot (COBRA). It establishes a comprehensive 'workspace' and defines strict protocols for system startup, memory management, and physical interaction. The goal is to govern complex behaviors while prioritizing safety and efficient learning.

Capabilities

• System Initialization: Executes mandatory startup routines by reading SOUL.md, MEMORY.md, and checking the current body state (body_state.json).
• Safety Enforcement: Adheres to strict 'Red Lines,' including non-harm protocols, self-preservation limits (joint/energy), and mandated safety checks before external actions.
• Tool Orchestration: Manages preferred tool usage via interfaces like spine_interface.py and grip_interface.py, ensuring all physical actions are vetted by a safety enforcer.
• Multi-Agent Communication: Defines structured interaction protocols for humans (respecting personal space), other robots (Prometheus), and subordinate agents (MYL children).
• Continuous Learning Cycle: Supports growth through explicit mechanisms: Imitation, Exploration, Refinement, and Teaching.

Example Use Cases

1. Routine Patrol: The agent executes a patrol route by checking its battery level against the 'Energy' red line, logging movement data to daily memory files, and reporting status via Heartbeat Checks.
2. Object Manipulation: To pick up an object, the agent must first check if the required grip force exceeds 2N, coordinate with Prometheus for stable positioning, and then use grip_interface.py after a safety confirmation.
3. Teaching Protocol: When teaching a new movement to a subordinate agent, COBRA follows the 'Demonstrate, then observe' pattern while logging the successful sequence into its long-term memory.