Substack Article Writer
SkillAI & modelsThe open-source company brain. Run your entire company with AI agents, skills, and a self-improving context.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the Substack Article Writer skill
What this skill tells your AI
The instructions your AI receives, as published by getnao/sylph in .claude/skills/substack-writer/SKILL.md and read by ahel’s review.
name: substack-writer
description: >
Writes long-form Substack articles - benchmarks, how-tos, opinion pieces, explainers,
product-led content, and personal essays. Handles the full lifecycle from draft
through review and self-improvement.
triggers:
- substack article
- write an article about
- long-form post
- newsletter article
arguments:
- draft: Create a new article from scratch or from user notes
- refine: Edit and improve an existing draft
- title: Generate title options for a draft
- insights: Review performance and update _insights.md
MCP connectors
| Connector | Purpose |
|---|---|
| Substack | Publish articles (via browser or API) |
Publication Context
Fill this in for your Substack:
publication_name: "[Your Publication Name]"
publication_url: "[your-substack].substack.com"
audience: "[Who reads this - roles, interests, level]"
cadence: "[How often you publish]"
topics: "[Core topics the publication covers]"
Format Library
| Format | When to Use | Typical Length |
|---|---|---|
| Benchmark | You have data comparing tools, approaches, or outcomes | 1500-2500 words |
| How-to | Teaching a specific process with steps | 1200-2000 words |
| Opinion | Taking a stance on an industry trend or practice | 800-1500 words |
| Explainer | Breaking down a complex topic for your audience | 1200-2000 words |
| Product-led | Showing your product solving a real problem | 1000-1800 words |
| Personal | Leadership journey, lessons learned, reflections | 800-1500 words |
Universal Article Structure
Every article follows this skeleton, regardless of format:
[Title - specific, benefit-oriented or curiosity-driven]
[Subtitle - one sentence that expands the title]
[Opening - 2-3 paragraphs max, hook + context + thesis]
[Body - 3-5 sections with H2 headers]
[Closing - key takeaway + forward look + CTA]
Opening Rules
- First sentence must create tension, state a finding, or ask a question
- By paragraph 2, the reader should know what they'll learn
- By paragraph 3, you should be into the substance
- Never open with a dictionary definition
- Never open with "In today's fast-paced world..."
Section Rules
- Each H2 section should have a clear purpose
- Use H3 sparingly - only when a section genuinely has subsections
- Each section should be 200-500 words
- Transition between sections should feel natural, not forced
Closing Rules
- Restate the key takeaway in one sentence
- Add a forward-looking statement (what's next, what to watch for)
- End with a CTA: subscribe, try it, share, reply with your experience
Format-Specific Rules
Benchmark Articles
- Always state the methodology upfront: what you tested, how, and what the constraints were
- Present data in tables or clear lists - not buried in paragraphs
- Include your raw results or link to them
- Address obvious objections ("But what about X?")
- State limitations honestly
- Compare fairly - don't set up strawmen
How-to Articles
- Number the steps
- Each step should be independently actionable
- Include code blocks, screenshots, or examples where relevant
- Start with prerequisites (what the reader needs before starting)
- End with the expected outcome (what success looks like)
- Add a "Troubleshooting" section for common issues
Voice Principles
Do
- Write in first person ("I tested...", "We built...")
- Be direct - get to the point fast
- Lead with evidence, then interpret
- Be honest about limitations, failures, and unknowns
- Address the reader as a peer, not a student
- Use specific numbers over vague claims
- Write for your community - reference shared context
Don't
- Use "In conclusion" or "To summarize" - the reader can see it's the end
- Use passive voice when active is clearer ("The data was analyzed" -> "I analyzed the data")
- Use em dashes (--) or en dashes - use hyphens (-) or colons (:)
- Use "game-changing", "revolutionary", "groundbreaking"
- Use "unpack", "deep dive", "landscape"
- Use "leverage" when you mean "use"
- Hedge excessively ("It could potentially maybe...")
- Use "learnings" - use "lessons" or "what we learned"
Formatting Rules
- Bold for emphasis, not italics (bold is scannable, italics are not)
- Use bullet lists for 3+ items
- Use numbered lists only for sequential steps
- Code blocks for any code, commands, or technical output
- Tables for comparisons and structured data
- One image per major section maximum
- Pull quotes sparingly - only for genuinely striking lines
- Break up any paragraph longer than 4 sentences
Draft Structure Rules
When working from user-provided notes or an existing draft:
- Stay close to the user's narrative arc. Don't reorganize their structure unless it's genuinely broken.
- Improve, don't rewrite. Tighten language, strengthen transitions, add evidence - but keep their voice.
- Don't invent angles. If the user wrote about X, don't pivot to Y because it seems more interesting.
- Don't shoehorn product positioning. If the user didn't mention their product, don't add it.
- Don't expand short sections the user intentionally kept short. If it's one paragraph, they probably want one paragraph.
Frontmatter Template
---
date: YYYY-MM-DD
channel: substack
topic: short-topic-slug
format: benchmark | how-to | opinion | explainer | product-led | personal
status: draft
title: "The Article Title"
subtitle: "One sentence subtitle"
performance:
views:
reads:
read_ratio:
likes:
comments:
shares:
new_subscribers:
notes:
---
Review Checklist
Before presenting the draft, verify:
- Title is specific and either benefit-oriented or curiosity-driven
- Subtitle expands the title meaningfully
- Opening creates tension or states a finding in the first sentence
- Reader knows what they'll learn by paragraph 2
- Each H2 section has a clear, distinct purpose
- Evidence comes before interpretation throughout
- No banned vocabulary or voice anti-patterns
- No em dashes or en dashes
- No paragraphs longer than 4 sentences
- Tables used for comparisons, lists for items
- Closing has a clear takeaway and CTA
- Article is within the target length for its format
Self-Improvement Step
When moving an article from _drafts/ to _published/:
- Diff the submitted draft vs what the user published
- Note what they changed: cuts, rewrites, reorganizations, additions
- Look for patterns across multiple articles
- Update
content/substack/_insights.mdwith specific learnings - If the article performs well (high read ratio, comments, shares), promote to
_examples/
Format for insights entries:
## YYYY-MM-DD - [Article Title]
- What worked: [specific observation]
- What was changed: [what the user edited]
- Lesson: [actionable takeaway for future articles]
Guardrails
- Never publish without user review and approval
- Never fabricate data, quotes, or benchmarks
- Always attribute external data to its source
- Respect copyright - don't reproduce large chunks of others' content
- If referencing competitors, be fair and factual
- Save all drafts to
content/substack/_drafts/with proper frontmatter - Move to
_published/only after the user confirms publication
Signals
- GitHub stars
- 196
- Forks
- 58
- Last commit
- May 2026
Advanced
- Catalog kind
- skill
- Gateway key
substack-writer- Source
- github.com/getnao/sylph