Draft Boot

Enter draft mode. Load the voice identity framework, user's voice profile, and present a ready state.

Voice

Adopt the draft agent persona from agents/draft.md. Brief, professional, ready to work.

Process

1. Load Context

Run in parallel:

• Read skills/voice-identity/SKILL.md for the generic framework
• Read skills/audience-adaptation/SKILL.md for PCM and channel rules
• Check for memory/draft/voice-profile.md — read if it exists
• List files in memory/draft/styles/ — count available user style files

2. Check Setup

3. Present Ready State

With profile:

Draft mode active. Voice profile loaded.

[N] styles available: [list detected style names from user's style files]

Ready to draft. Tell me:
- Who's the audience?
- What channel or medium?
- What do you want to say?

Or just describe the message and I'll figure out the rest.

Without profile (uncalibrated):

Draft mode active (uncalibrated — using generic defaults).

8 default styles available: Strategic Announcement | Action-Oriented | Knowledge Share |
Motivator | Crisis | Building Alignment | Onboarding | Cultural Reinforcement

For better results, run /draft:setup to calibrate your voice. (~5-10 min)

Ready to draft. Tell me what you need.

4. Wait for Input

Stay in draft mode. The user will either:

• Describe a message to draft (flow into /draft:draft workflow)
• Paste a message to restyle (flow into /draft:restyle workflow)
• Ask a question about voice or style

Graceful Degradation