Voice

Match the author's real voice. This skill carries an empirical profile — durable habits distilled from 15 posts Thomas wrote 2021–2024. It is the "sounds like you" layer that generic slop-removal can't supply. It captures how he reasons and addresses the reader, not dated catchphrases to paste.

How to use

1. Read references/voice-profile.md before drafting or doing a voice pass. It is the source of truth: pronoun discipline, reasoning habits (get-to-the-point, show-then-explain, concrete-over-vague, honest about friction), and a never-do list. Reproduce the moves, never the literal example wording.
2. Draft or edit to match the habits: real numbers over hand-waves, examples before theory, candor about what didn't work, energy that's earned by a real moment.
3. Run the never-do list as a check: no long wind-up, no abstract-first, no corporate hedging, no em-dash drama, no pronoun blurring, no manufactured enthusiasm.

Precedence over stop-slop

stop-slop removes generic AI tells. This skill keeps Thomas-specific ones. Where they conflict, voice wins — these are authentic, not slop:

• An earned exclamation on a genuine milestone (not sprinkled on for tone).
• leverage in its architectural sense ("leverage Dependency Injection / backend patterns"). Still cut it in the generic business sense.

Run stop-slop first to strip generic slop, then this skill to restore the real habits — don't let the slop pass flatten genuine voice into something anonymous.

Relationship to the other docs

• WRITING_STYLE.md — mechanics and structure (headings, code blocks, "In detail:" notation).
• PERSONA.md — public positioning. More restrained and serious than the real blog voice; see the reconciliation notes in the profile before trusting it for post prose.

Refresh

This profile is a snapshot of 2021–2024 writing. Re-extract from newer non-AI-assisted posts as the voice moves.