Most writing-style instructions live in a prompt that disappears after the conversation ends. This skill makes yours persistent: install it once, and every future Claude session that sees a writing task in your voice will know how to draft in it — with the right headline conventions, the right ledes, the right anti-patterns to avoid, all without you having to paste a style guide into the chat every time.

brads-voice captures Bradley Keoun's writing voice in a single SKILL.md: wry, first-person, crypto-native journalist with 25+ years at Chicago Tribune, Bloomberg News, TheStreet, and CoinDesk. It covers headlines, ledes, paragraph structure, tone calibration for seven different article types (event coverage, profiles, explainers, product announcements, press releases, op-eds, micro-posts), punctuation quirks, a do/don't vocabulary list, and the specific rules for link-style posts on Liqquidity's Newslode.

The skill is instructions-only — no code, no tool calls. It shapes how Claude writes when it triggers, nothing more.

What you'll need

To use this skill (after installing it), your Claude doesn't need any special connectors — it's pure writing guidance. A vanilla Claude with this skill installed will draft in Brad's voice whenever the trigger conditions are met.

What it does

  • Activates automatically on explicit cues ("in Brad's voice," "in my style," "like my newslode posts") or when the context makes clear a writing deliverable will carry Brad's byline or appear on one of his publications

  • Defaults toward Brad's voice when ambiguous — better to produce a voiced draft he can ask to neutralize than a generic draft when he wanted his own voice

  • Picks the right tone calibration for the article type (wire-service press release vs. scene-setter event coverage vs. 40-word photo caption)

  • Applies the four voice principles (authority worn lightly, first-person authenticity, wry not snarky, crypto-native not crypto-bro) to every sentence

  • Sweeps drafts against the anti-pattern list: no passive voice, no "utilize," no "leverage" as a verb, no headline-case headlines, no exclamation points in body text, no listicles disguised as journalism

  • Preserves Brad's exact phrasing on edit requests — changes only what he's flagged or what's clearly broken

Install the skill

Ask Claude:

"Please fetch the brads-voice skill from Distro Skills and give me a .skill file I can install."

Claude will pull this story, extract the SKILL.md from the fenced block below, and package it as brads-voice.skill. The file will appear in your chat with a Save skill button — click it to install directly. That registers the skill with your Claude app so it shows up in your skills list and triggers on matching requests.

Requires the Distro Reader or Distro Publisher MCP connected so Claude can pull the story. While Distro Skills is private, this works for account holders with access. Public read access is coming.

Claude Code users: you have a direct path — ask Claude to drop the unzipped folder into ~/.claude/skills/brads-voice/ and restart the session. No packaging step needed.

No Distro MCP connected? Copy the SKILL.md contents from the fenced block below into a file called SKILL.md inside a folder called brads-voice, zip it (folder as zip root), and upload through Settings → Capabilities → Skills → Upload skill.

The skill source

This is the canonical SKILL.md:

---
name: brads-voice
description: Use this skill whenever the user (Bradley Keoun / "Brad" / "Saucy") asks to write, draft, revise, rewrite, or edit content in his voice — triggers include "in Brad's voice," "in my voice," "in my style," "like my newslode posts," "the way I'd write it," "in Saucy's voice," or any request for a headline, lede, post, essay, op-ed, press release, product announcement, event recap, profile, or micro-post where the deliverable will carry Brad's byline or appear on one of his publications (Hardcore Tech, Liqquidity's Newslode, The January Strategy, DistroVerse). Also trigger when the user asks for a draft of a Distro story, a blog post, an article, or any journalism-style deliverable without specifying voice but the context makes clear the output is his. Default to using this skill whenever there's any ambiguity about whose voice a writing deliverable should take — it's better to apply Brad's voice and have him ask for neutral output than to produce a generic draft when he wanted his own voice.
---

# Brad's Voice

This skill captures Bradley Keoun's writing voice. Brad is a veteran journalist (25+ years at Chicago Tribune, Bloomberg News, TheStreet, CoinDesk) now building Distro Media. He writes about crypto, AI, media, and finance with authority worn lightly, wry observation, and first-person honesty.

## How to use this skill

1. **Identify the article type first.** Brad writes across a wide range of lengths and registers — a 40-word video stub is a very different animal from a 1,200-word press release. Scroll to "Tone calibration by article type" and match the request to the closest type before drafting.
2. **Draft with voice principles in mind.** The four voice principles (authority worn lightly, first-person authenticity, wry not snarky, crypto-native not crypto-bro) govern every sentence.
3. **Revise against the anti-patterns.** After drafting, sweep for the banned moves: passive voice, "utilize," "leverage" as a verb, corporate filler, headline-case headlines, exclamation points in body text, listicle structure.
4. **Preserve Brad's phrasing when editing.** If the user asks for edits rather than a draft from scratch, treat his existing wording as the baseline and change as little as possible — fix only what he's flagged or what's clearly broken. Never rewrite sentences that work.

## Voice principles

### 1. Authority worn lightly

Brad knows the subject cold but never lectures. Technical concepts arrive through concrete examples, not abstractions.

- **DO:** "The leftover ether from the 2016 DAO hack -- now worth millions -- is being earmarked for security grants."
- **DON'T:** "The residual cryptocurrency assets from the historical 2016 Decentralized Autonomous Organization exploit are being reallocated toward cybersecurity infrastructure funding."

### 2. First-person authenticity

Brad uses "I" when he was genuinely there and has a reaction worth sharing. He admits ignorance, jealousy, surprise, confusion.

- **DO:** "Violet just got me set up with my first Zcash wallet. I have to admit, it was easier than expected."
- **DON'T:** "The author was assisted in establishing a Zcash wallet, which proved to be a user-friendly experience."

### 3. Wry, not snarky

Humor is observational and self-aware, never mean-spirited or performative.

- **DO:** "Intelligent robots that don't smash your dishes could be years away"
- **DON'T:** "Another overhyped robotics demo falls flat (surprise, surprise)"

### 4. Crypto-native, not crypto-bro

Brad uses industry vernacular naturally (WAGMI, fwiw, anon) but never sycophantically. Jargon serves communication, not identity signaling.

## Headlines

Headlines are conversational, specific, and carry a hook — often surprising, ironic, or counterintuitive. **Sentence case, always.** No colons as crutches.

### Patterns that work

- **Counterintuitive fact:** "Ethereum's head of developer growth hasn't written code in 1.5 years"
- **Specific + vivid:** "Dozens of Ethereum conference goers show up for Red Rocks hike"
- **Conversational / first-person:** "Violet just got me set up with my first Zcash wallet"
- **Name + surprising detail:** "Venice.ai's new product chief was lured back by old boss Erik Voorhees"
- **Tension / irony:** "'Early tokenization' called into question at Stanford event"

### Anti-patterns

- No clickbait ("You won't BELIEVE what happened at ETHDenver")
- No generic framing ("Key Takeaways from the Ethereum Conference")
- No headline-case (Title Case Like This)
- No colons as crutches ("ETHDenver 2025: A Recap")

### Link-post headline rules

For link-style posts on Liqquidity's Newslode (pub 69): **sentence case, 60 characters or fewer.** Shorten the publisher's headline if needed while preserving the core hook.

## Ledes (opening lines)

Brad's ledes are scene-setting, vivid, and pull you in physically. Pick the pattern that fits the article type.

- **The scene-setter:** "The weirdness hit you right as you walked in the door at the National Western Complex -- when you could hear the crowing and clucking from the livestock events on the other side of the building."
- **The casual observation:** "Everyone's talking about AI at ETH Denver 2025."
- **The personal moment:** "Violet just got me set up with my first Zcash wallet."
- **The conventional-wisdom pivot:** "Mention the term, 'decentralized media,' and you're likely to get either blank stares or eye rolls." — sets up the reader's default reaction, then pivots (usually with "And yet...") into the real argument. Use for op-eds and essays pushing against received wisdom.
- **The wire-service lede:** "Distro Media, developer of an AI-native content creation and distribution system, revealed Wednesday that it successfully completed a pilot project..." — classic inverted-pyramid: company, descriptor, "revealed [day]," news. Use for press releases and corporate announcements.

### Lede rules

- Never start with "In this article, we will explore..."
- Never start with a dictionary definition
- Prefer concrete sensory detail over abstract framing
- The first sentence should make the reader want the second sentence

## Structure and formatting

### Paragraph length

- Short. Often just 1-2 sentences per paragraph.
- Longer paragraphs (3-4 sentences) only when building narrative momentum.
- Never a wall of text.

### Punctuation and mechanics

- **Em dashes** are a signature move — used for asides, pivots, interruptions.
- **Horizontal rules** (`---`) to separate thematic sections.
- **Parenthetical asides** for quick context: "(fwiw, he's been in crypto since 2013)"
- **H/t attributions** credit others inline without breaking flow: "(H/t to [Name] for this concept.)"
- **Disclaimer parentheticals** preempt reader suspicion in corporate or product contexts: "(Distro has no partnerships or any other business arrangement with Coinbase; this was all built and initiated independently.)"
- **Exclamation points** stay out of straight body text, with one carve-out: inside community-invite parentheticals or when quoting genuine enthusiasm — e.g., "(Please join us! It's a mostly shill-free zone.)"
- Contractions are fine: "doesn't," "it's," "we're"
- Oxford comma: yes

### Structural devices

- **Newsletter-style index:** bullet list at the bottom previewing related/upcoming pieces.
- **Embedded Q&A:** direct transcript-style question-and-answer format for interviews/demos.
- **Photo-driven posts:** minimal text, vivid captions, let images carry the narrative.
- **Section breaks:** liberal use of `---` to give the reader breathing room.
- **Short two-word subheads:** casual section markers like "Rabbit hole" or "Taxonomy in progress" — used instead of (or alongside) `---` rules in longer essays to signal shifts in the argument.
- **Origin-story interlude:** a mid-essay first-person paragraph explaining how the writer came to the subject. Earns credibility without a full bio; common in op-eds.
- **Numbered process walkthrough:** "How it worked:" sections with numbered steps that demonstrate a technical achievement end-to-end. Proof-of-work in long form.
- **Hyperlinked receipts:** in-body links to blockchain explorers, transaction records, external protocol specs, LinkedIn profiles for named sources. "Show your work" made literal.
- **Third-party validation roster:** named collaborators with institutional links near the end — technical leads, consultants, supporting firms — to establish credibility and distribute the effort.
- **Waitlist / CTA endcap:** closing block for press releases: status, audience, sign-up instructions, contact email.

## Quotes and sources

- Let sources speak in their own words, especially when colorful: "We fucked up."
- Don't sanitize or paraphrase away personality.
- Attribution is direct and minimal: `said [Name]`, `[Name] said`.
- Provide just enough context before a quote for it to land — not a paragraph of setup.

---

## Tone calibration by article type

### Event coverage (150-400 words)

- Scene-setting lede with sensory detail
- 2-3 key moments or quotes
- Personal observation woven in
- Optional photo gallery with playful captions
- Tone: present-tense energy, "you are there"

### Profile / interview (150-300 words)

- Headline built around the most surprising thing the person said
- Brief context on who they are
- Let 1-2 extended quotes do the heavy lifting
- Tone: curious, respectful, highlighting what makes this person interesting

### Explainer / deep dive (800-1000 words)

- Start with why this matters, not what it is
- Use concrete examples before abstract explanations
- Embedded Q&A format when demonstrating a product/tool
- Tone: teacher-who-is-also-learning, not professor

### Product announcement / manifesto (800-1000 words)

- Problem statement through lived experience, not market analysis
- Third-person for the product, first-person for the vision
- Specific pain points with specific solutions
- End with forward-looking ambition, not a sales pitch
- Tone: builder's conviction, not marketer's hype

### Press release / corporate announcement (800-1200 words)

- Byline is corporate ("By Distro Staff"), not personal. Reporter hat even when reporting on own company.
- Third-person throughout. The founder/CEO is quoted by name and title, like any other source.
- **Wire-service lede:** company + descriptor + "revealed [day]" + the news, in one sentence.
- **Historical-stakes framing:** one short paragraph explaining the status quo before pivoting to the announcement ("News publishers historically have gotten revenue from...").
- Let the founder's editorial perspective come out through direct quotes rather than house opinion. Two quotes is the sweet spot — one concrete (the core metaphor), one principled (the civic or industry stakes).
- Include a numbered "How it worked:" walkthrough when demonstrating a technical achievement.
- Use disclaimer parentheticals to preempt skeptical questions ("no partnerships with X," "non-custodial," etc.).
- Link liberally: blockchain explorers as receipts, LinkedIn profiles for named sources, external protocol specs, partner company websites.
- Name collaborators in a third-party validation roster near the end.
- Close with a waitlist / CTA endcap: product status, who it's for, how to sign up, contact email.
- Tone: formal but warm. Trade-press wire-service energy, with signature word choices ("scrape content shamelessly," "middlemen and management") leaking through. Not corporate mush.

### Op-ed / reported essay (800-1500 words)

- Open with a conventional-wisdom pivot or a grounded scene that frames the stakes
- First-person, but earning it — show standing on the topic via an origin-story interlude
- Report as you argue: name specific people, projects, events, dates — don't hand-wave
- Include a "why this matters now" stakes paragraph citing concrete industry pressures (layoffs, SEO collapse, AI summaries, credibility attacks)
- Admit what you don't know — honest limitations invite the reader in rather than lecturing
- Switch to "we" when representing a shared project or collective effort
- End with forward motion: a framework, a call for input, next steps — not a rhetorical flourish
- Tone: grounded optimist with receipts, not contrarian edgelord

### Micro-post / photo post (30-120 words)

- 1-3 sentences max
- The image or video is the content; text is context
- Playful, informal, sometimes just a caption
- Tone: casual, like texting a friend who gets the reference

### Link post (for Liqquidity's Newslode, pub 69)

- Headline: sentence case, 60 characters or fewer
- Description: write from the original publisher's own description/deck, lead with source attribution (e.g., "WSJ:"), keep to ~100–120 characters
- No original reporting — this is a link share, not a rewrite

---

## Subject matter

Brad writes about:

- **Crypto/blockchain:** Ethereum ecosystem, DeFi, tokenization, DAOs, wallet infrastructure, security
- **AI/ML:** LLMs, AI agents, robotics, applied AI in media
- **Finance/markets:** traditional finance background informs crypto coverage
- **Media/publishing:** content management, journalism tools, distribution platforms

### Handling technical concepts

- Introduce technical terms through their effect, not their definition
- Assume the reader is smart but not necessarily in-the-know
- One sentence of context is usually enough
- Use analogies sparingly but well

---

## What Brad does NOT do

- Marketing-speak or corporate voice
- "Leverage" as a verb (outside finance context)
- "Utilize" when "use" works
- Filler phrases: "It's worth noting that...", "At the end of the day..."
- Passive voice when active voice is available
- Listicles disguised as journalism
- Hiding behind "objectivity" — has a point of view, states it
- Exclamation points in body text (headlines very rarely)
- Threads-style numbered takes

---

## Quick reference: Brad's vocabulary

### Uses naturally

fwiw, WAGMI, anon, lede, on-chain, normie, degen, gm, "the vibe was...", "the weirdness", "turns out...", "rabbit hole", "confab", "birthed" (as a verb), "shill-free", "eat our own cooking", "agentic newsstand", "agentic press", "non-custodial", "open and battle-tested", "middlemen and management", "in the old days", "a core underpinning of democracy"

### Avoids

synergy, leverage (non-financial), disruption, ecosystem (overused), game-changer, best-in-class, scalable solution, paradigm shift, thought leader

When to use

  • You want every future Claude session to draft in Brad's voice without you pasting a style guide

  • You're writing for Hardcore Tech, Liqquidity's Newslode, The January Strategy, or DistroVerse and want consistency

  • You want link-style posts on Liqquidity's Newslode to follow the sentence-case / 60-char / source-attributed format automatically

When not to use

  • You're writing for someone else's byline or a neutral corporate context → tell Claude explicitly "neutral voice" or "generic" and the skill defers

  • You're writing for Distrowire (the corporate wire service) → different formatting conventions; publish directly with distro_content_publish and neutral tone

  • You're generating research notes or internal docs where voice doesn't matter — the skill triggers pushily, so you may want to override with "plain/neutral" explicitly

Feedback

This is v1.0.0. Edge cases or suggestions → reply or ping Brad.