Add Development Constitution (`copilot-instructions.md`) with Agent-Embedded Skills

[tmcclell]

Add Development Constitution (copilot-instructions.md) with Agent-Embedded Skills

Problem

ODrive lacks a copilot-instructions.md file. When developers start new Copilot chat sessions, there's no baseline instruction to establish project values, principles, or guidance on available agents/capabilities.

Proposed Solution

Two-part architecture:

1. copilot-instructions.md — Development constitution + agent registry

   • Core principles and values (300-400 words)
   • How to use Copilot effectively
   • Directory of available agents with descriptions
   • Single entry point for new chats

2. Agent files with embedded Skills sections

   • Each agent includes a ## Skills section containing task-specific context
   • Self-contained: agent definition + tactical guidance in one file
   • Auto-loaded when agent is invoked (e.g., @firmware-engineer)
   • No separate primer files; everything is in .github/agents/

Acceptance Criteria

Constitution File

• Create .github/instructions/copilot-instructions.md
• Include core principles (~300-400 words)
• Include "How to use Copilot" section
• Include agent registry with descriptions
• Reference agent names (e.g., @firmware-engineer) users can invoke
• Verify file loads cleanly in Copilot chat (~1-2K tokens)

Agent Skills Sections

Update or create agents with embedded Skills sections:

• workspace.agent.md: Add "Workspace Development Context" skill
• project-lead.agent.md: Add "Project Orchestration Context" skill
• firmware-engineer.agent.md: Add "Firmware Development Context" skill
• qa-engineer.agent.md: Add "QA & Testing Context" skill (create if missing)
• hardware-engineer.agent.md: Add "Hardware Design Context" skill
• motor-control-engineer.agent.md: Add "Motor Control Context" skill

Each Skills Section Should Include

• Current architecture overview
• Key constraints and requirements
• Core responsibilities
• Key files and responsibilities
• Relevant code patterns or examples
• Timing/performance expectations (if applicable)
• Real-time or domain-specific considerations

Documentation

• Update README or CONTRIBUTING.md:
  • Link to copilot-instructions.md for new developers
  • Explain how to invoke agents (@agent-name or /slash-command)
• Add note: "Load copilot-instructions.md in new chat, then use available agents"

Testing

Test Case	Action	Expected Outcome
Constitution Loading	Load copilot-instructions.md in new Copilot chat	File is clear, actionable, and under 2K tokens
Agent Invocation	Invoke agent (e.g., @firmware-engineer)	Skills section loads; agent persona and context are present
Slash Command	Test invocation via slash command (e.g., /firmware-optimize)	Slash command resolves to agent; agent + skills appear in response
Context Switching	Switch between agents mid-chat	Context adapts correctly; previous agent context doesn't bleed into new agent