From thinking-frameworks-skills
Writes 60-140 word third-person blurbs for Substack cross-posts, ready for direct paste by other newsletter writers without editing. Triggers on cross-post, blurb keywords for positioning tech essays.
npx claudepluginhub lyndonkl/claude --plugin thinking-frameworks-skillsThis skill uses the workspace's default tool permissions.
```
Generates design tokens/docs from CSS/Tailwind/styled-components codebases, audits visual consistency across 10 dimensions, detects AI slop in UI.
Records polished WebM UI demo videos of web apps using Playwright with cursor overlay, natural pacing, and three-phase scripting. Activates for demo, walkthrough, screen recording, or tutorial requests.
Delivers idiomatic Kotlin patterns for null safety, immutability, sealed classes, coroutines, Flows, extensions, DSL builders, and Gradle DSL. Use when writing, reviewing, refactoring, or designing Kotlin code.
Write blurb for cross-post:
- [ ] Step 1: Load spine + voice-profile + audience-notes
- [ ] Step 2: One-line positioning: "In this piece, Kushal argues…"
- [ ] Step 3: 2-3 sentences summarizing argument + one concrete anchor
- [ ] Step 4: One sentence on why a tech-adjacent audience should care
- [ ] Step 5: Sign-off minimal: "Read the full piece." or a period — no subscribe CTA
- [ ] Step 6: Enforce 60-140 words
---
source_post: {slug}.md
platform: substack-crosspost
target_length: 60-140 words
actual_length: {N}
section: {section-slug}
---
In this piece, Kushal argues {thesis}. {2-3 sentences of argument + concrete anchor}. For anyone building with {domain}, the {specific-claim} is the move that reframes the problem. Read the full piece.
---
source_post: architecture-not-prompting.md
platform: substack-crosspost
target_length: 60-140 words
actual_length: 98
section: agent-workshop
---
In this piece, Kushal argues that most prompt-engineering advice mistakes the unit of analysis — the lever that moves behaviour in long-running AI systems is organisational, not linguistic. Drawing on Wang et al., Anthropic (2024), which found multi-agent decompositions outperformed long-prompted single agents by roughly 40% on sustained tasks, he lays out four recurring patterns: supervisor-worker, pipeline, jury, and debate. For anyone building with agents — or watching teams ship agents that don't quite work — the reframing from "better prompt" to "better architecture" is worth the ten-minute read.