Write Like Me

[!IMPORTANT] Scope Directive This is the unified facade and orchestrator for authentic user writing voice emulation and profile maintenance. It routes between runtime drafting and administrative voice profile management. It owns voice emulation, style calibration, and learning from approved edits. It does NOT own filing drafted notes into a PARA vault (delegated to para-vault) or raw filesystem mechanics (delegated to manage-file-operations).

---

Triage Matrix: Selecting the Action Runbook

Match the user's intent to the appropriate workflow:

---

Core Guidelines & Invariants

1. Facts & Intent Preservation: Emulating voice never authorizes changing supplied facts, commitments, constraints, or uncertainty levels.
2. Drafting Reads; Maintenance Writes: Ordinary drafting reads profile files and leaves their bytes unchanged. Profile updates occur only when explicitly requested or when draft corrections are approved for learning.
3. Evidence-Backed Calibration: Base tone guidelines strictly on authentic user samples, never on agent hallucinations or third-party boilerplate.
4. Scope Precedence: Resolve configurations in canonical order: explicit pointer $\rightarrow$ project override (.nyssa-ai/agent-skills/writing-voice/) $\rightarrow$ user fallback (~/.nyssa-ai/agent-skills/writing-voice/).

---

Verification

Before reporting completion of any writing voice task:

• For drafting tasks, confirm facts and meaning match the source request, with personal tone applied without altering commitments.
• For profile updates, verify modified files are saved in the authorized user or project scope.
• Confirm no private scratch or confidential sample text leaks into persistent public outputs.
• Confirm VOICE.md and STYLE.md conform to the profile schema in references/profile-schema.md.