Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Transform brain dumps into polished blog posts in Nick Nisi's voice. Use when the user wants to write a blog post with scattered ideas, talking points, and conclusions that need organization into a cohesive narrative with Nick's conversational, authentic, and thoughtful tone.
.claude/skills/microck-blog-post-writer/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 74% | 0% |
| case-10 | ✗→✓ | ▲ Improved | 64% | 0% |
| case-20 | ✓→✗ | ▼ Worse | -19% | 0% |
| case-21 | ✓→✗ | ▼ Worse | 26% | 0% |
| case-02 | ✓→✓ | = Same ✓ | 77% | 0% |
Transform unstructured brain dumps into polished blog posts that sound like Nick Nisi.
Accept whatever the user provides:
Don't require organization. The mess is the input.
Load references/voice-tone.md to understand Nick's writing style.
Key characteristics:
Read references/story-circle.md to understand the narrative framework.
Determine if the content fits a story structure:
Not every post needs the full Story Circle, but look for narrative opportunities.
Structure the material into sections:
Common structures:
Choose the structure that fits the content.
Apply voice characteristics:
Opening:
Body:
Technical content:
Tone modulation:
Ending:
Check the post:
Show the post to the user for feedback and iterate.
markdown"AI is going to replace developers." I must have heard that phrase a hundred times in the last year.
markdownI've been thinking a lot about how we use AI in our daily work.
markdownThen something clicked. I watched it use rg to search through codebases, just like I would.
markdownI won't lie – joining Meta was intimidating.
markdownI watched it use `rg` to search through codebases, just like I would. It ran `npm test` to verify its changes weren't breaking anything.
markdownYou're not being replaced; you're being amplified.
references/voice-tone.md - Complete voice and tone guide. Read this first to capture Nick's style.references/story-circle.md - Story Circle narrative framework. Check if content fits a story structure.User provides brain dump:
thoughts on using cursor vs claude code
- cursor is in IDE, feels familiar
- but claude code is in terminal, my natural environment
- tried cursor first, felt weird leaving vim
- claude code met me where I was
- not about which is better, about workflow fit
- some devs love IDE integration
- I need terminal access
- conclusion: use what fits YOUR workflowProcess:
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | pass→pass | 12,234 | 15,481 | +27% | 1 | 1 | 0% | 1,838 | 3,259 | +77% | 0 | 0 | — |
case-15 | pass→pass | 13,995 | 10,777 | -23% | 1 | 1 | 0% | 2,090 | 2,625 | +26% | 0 | 0 | — |
case-01 | fail→pass | 11,818 | 13,420 | +14% | 1 | 1 | 0% | 1,804 | 3,135 | +74% | 0 | 0 | — |
case-03 | pass→pass | 14,602 | 15,774 | +8% | 1 | 1 | 0% | 2,181 | 3,658 | +68% | 0 | 0 | — |
case-04 | pass→pass | 11,715 | 11,885 | +1% | 1 | 1 | 0% | 1,842 | 2,846 | +55% | 0 | 0 | — |
case-05 | pass→pass | 13,771 | 13,063 | -5% | 1 | 1 | 0% | 1,960 | 2,975 | +52% | 0 | 0 | — |
case-06 | pass→pass | 15,821 | 13,107 | -17% | 1 | 1 | 0% | 2,409 | 3,135 | +30% | 0 | 0 | — |
case-07 | pass→pass | 13,290 | 13,056 | -2% | 1 | 1 | 0% | 2,125 | 3,084 | +45% | 0 | 0 | — |
case-08 | pass→pass | 10,892 | 13,911 | +28% | 1 | 1 | 0% | 1,636 | 3,093 | +89% | 0 | 0 | — |
case-09 | pass→pass | 14,922 | 13,112 | -12% | 1 | 1 | 0% | 2,387 | 3,097 | +30% | 0 | 0 | — |
case-10 | fail→pass | 11,353 | 12,096 | +7% | 1 | 1 | 0% | 1,916 | 3,150 | +64% | 0 | 0 | — |
case-11 | pass→pass | 9,192 | 13,172 | +43% | 1 | 1 | 0% | 1,379 | 3,058 | +122% | 0 | 0 | — |
case-12 | pass→pass | 15,981 | 15,876 | -1% | 1 | 1 | 0% | 2,234 | 3,450 | +54% | 0 | 0 | — |
case-13 | pass→pass | 14,142 | 11,815 | -16% | 1 | 1 | 0% | 2,046 | 2,786 | +36% | 0 | 0 | — |
case-14 | pass→pass | 17,884 | 14,446 | -19% | 1 | 1 | 0% | 2,722 | 3,293 | +21% | 0 | 0 | — |
case-16 | pass→pass | 9,998 | 13,743 | +37% | 1 | 1 | 0% | 1,642 | 3,203 | +95% | 0 | 0 | — |
case-17 | fail→fail | 14,652 | 15,878 | +8% | 1 | 1 | 0% | 2,258 | 3,520 | +56% | 0 | 0 | — |
case-18 | pass→pass | 13,705 | 12,188 | -11% | 1 | 1 | 0% | 2,261 | 2,924 | +29% | 0 | 0 | — |
case-19 | pass→pass | 9,059 | 12,915 | +43% | 1 | 1 | 0% | 1,368 | 2,889 | +111% | 0 | 0 | — |
case-20 | pass→fail | 26,762 | 16,385 | -39% | 1 | 1 | 0% | 4,213 | 3,423 | -19% | 0 | 0 | — |
case-21 | pass→fail | 16,484 | 16,318 | -1% | 1 | 1 | 0% | 3,268 | 4,122 | +26% | 0 | 0 | — |
case-22 | pass→pass | 21,056 | 17,542 | -17% | 1 | 1 | 0% | 3,191 | 3,662 | +15% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of 0 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.