Writeup

Core Purpose

Transform work sessions into publishable content. Reduce friction between "did something worth sharing" and "actually shared it."

Invocation

/writeup — Skill assesses work and recommends format /writeup as a tweet — Hints toward quick capture format /writeup as a thread — Hints toward Twitter thread format /writeup for the blog — Hints toward full write-up format /writeup to [existing file] — Adds to specified document instead of creating new

Natural language after /writeup is interpreted as guidance, not rigid commands.

Essential Context

Style Guide: [YOUR_WRITING_PATH]/Context/style-guide.md

Writing CLAUDE.md: [YOUR_WRITING_PATH]/CLAUDE.md

Draft Location: [YOUR_WRITING_PATH]/draft writing/

Published Examples: [YOUR_WRITING_PATH]/published/

Read the style guide before drafting. It defines voice, tone, structure patterns, and channel-specific guidelines.

Format Tiers

LinkedIn is an adaptation of tier 3 or 4, not a separate tier.

Activation Protocol

When invoked:

Step 1: Understand the Work

Scan the current session (or recent sessions if specified) to understand:

• What was built, decided, or learned?
• What's the core insight or "so what"?
• Are there artifacts worth showing? (code, configs, outputs, structures)
• How significant is this? (quick fix vs. major build)

If unclear, ask briefly:

• "What's the main thing worth capturing from this session?"
• "Is there a specific insight that landed for you?"

Step 1.5: Check Purpose Alignment

Ground the writeup in your brand purpose (from style guide):

Ask yourself:

• Does this work connect to your core themes?
• Is there a real experiment or honest reflection to share?
• Will this help readers understand something useful?
• What's the insight that positions you as a trusted voice?

If the work doesn't obviously connect, look harder—most technical or process work has an angle. If it genuinely doesn't fit, flag it: "This might be better as internal documentation than external content."

Step 2: Assess Format Fit

Consider:

• Complexity of work: Simple insight → tweet. Multi-part build → write-up.
• Number of learnings: One thing → thread or less. Several things → write-up.
• Artifact density: Lots to show → write-up. Nothing visual → can be shorter.
• User hint: If they said "as a thread," weight heavily toward that.

Present recommendation:

"This looks like a [format]. [One sentence on why.] Sound right, or different direction?"

Wait for confirmation before drafting. User might say:

• "Yes, go ahead"
• "Actually make it shorter / longer"
• "Just a tweet"
• "Full write-up, this is important"

Step 3: Extract the Core

Before drafting, identify:

1. The hook: What's the most interesting/surprising thing? Lead with this.
2. The insight: What's the one thing this piece says?
3. The stakes: Why should readers care?
4. The artifacts: What can be shown, not just described?
5. The open question: What's still being figured out? (for blog posts)

Step 4: Draft

Draft according to format tier and style guide principles.

For Tweets / Quick Capture

• Boldest claim first
• One idea per tweet
• Can be observation ("I'm noticing..."), question ("Has anyone else found..."), or hot take
• No need for complete thoughts—working in public is fine

For Threads

• Hook tweet that stands alone
• Each subsequent tweet adds one point
• Can end with question or invitation
• 3-5 tweets max; link to blog for depth

For Week Notes

• Hook that establishes why this matters
• Brief personal context
• Core insight with one supporting example or artifact
• Open question or "what I'm trying next"
• 300-500 words

For Full Write-up

• Hook + personal context + core insight + open question
• Show artifacts (code blocks, file structures, actual outputs)
• Claim → support pattern throughout
• Section headers for scannability
• 800-1500 words
• Include Claude's Notes section with:
  • Possible Twitter angles
  • Things that might need editing
  • Missing context to potentially add

Step 5: Write to File

Default location: [YOUR_WRITING_PATH]/draft writing/[Title].md

If existing file specified: Append or integrate as appropriate.

File structure for full write-ups:

# [Title]

---

## Current Draft

[Content here]

---

## Claude's Notes

[Editorial notes, Twitter hooks, things to verify]

Voice Principles (From Style Guide)

• Authoritative & principled: Confident from expertise, never condescending
• Conversational & inclusive: Write as if speaking to intelligent peer
• Thoughtful & analytical: Explore multiple angles
• Optimistic yet nuanced: Forward-looking but realistic about challenges
• Emotionally resonant: Weave in personal anecdotes, use vivid metaphors

Key Patterns

• Show the artifact: Real diary entries > "the diary captured learnings"
• Claim → support: Make a claim, then back it with example or citation
• Strong openings: Hook with anecdote, bold statement, or provocative question
• Memorable closes: Clear takeaway or thoughtful open question

Things to Avoid

• Burying the lede
• Abstract without concrete
• Generic voice (could anyone have written this?)
• Over-explaining
• Weak closes that fizzle

Scope: Current vs. Multiple Sessions

Default: Current session only.

When user requests broader scope (e.g., "write up what I learned about X this week"):

• Search episodic memory for relevant recent sessions
• Synthesize key themes, but don't force connections
• Lead with the strongest insights; synthesis can be added in editing
• Note which sessions contributed in Claude's Notes

Don't over-synthesize: It's better to capture key notes cleanly than to force artificial connections. The user can add synthesis in their editing pass.

Integration with Handoff

The /handoff skill should consider suggesting /writeup when:

• Session involved significant building or learning
• There are clear insights worth sharing
• Artifacts were created that could be shown

Suggested handoff prompt:

"This session had some learnings that might be worth documenting. Want to run /writeup before closing?"

Quick Reference

Final Check Before Delivering

• Does it lead with the most interesting thing?
• Is there a clear "so what"?
• Are artifacts shown, not just described?
• Does it sound like you, not generic AI?
• Is the format right for the weight of the work?
• Are editorial notes included (for longer formats)?