shorts
Use to convert any input into a ready-to-record YouTube Shorts script.
Skill by aayushostwal
Install
Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.
npx agentmarket install skill/aayushostwal-shorts
Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.
Get the weekly diff
The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.
About
Use cases: writing
Reviews
Write a reviewNo reviews yet. Installed it? Reviews come from verified installers only. How reviews work
Permissions
What this skill can do and what it asks for, from its SKILL.md and code. Checked Oct 25, 2026.
- Tools
- No tools found
- Secrets
- None asked for
- Install scripts
- None found
Network hosts it talks to 1
codeload.github.com· found in install command
Files it touches
~/.claude/skills/shorts
What's inside
- Skill name
shorts- Description
- Use to convert any input into a ready-to-record YouTube Shorts script. Trigger on requests for 30-second scripts, hooks, reels/short-form video copy, or pasted topics/articles/bullets/URLs needing concise high-retention narration with CTA.
- License
- MIT
- Instructions
- 201 lines in SKILL.md · view on GitHub
Files (4)
- SKILL.md
- common-mistakes.md · anti-patterns/
- execution-checklist.md · checklists/
- output-validation.md · validation/
SKILL.md (first lines)
# Nexus YouTube Shorts Script Writer Turn any input into a ready-to-record 30-second YouTube Shorts script: hook, core content, and CTA. --- ## Compatibility - Input: any format — topic keyword, bullet points, article text, URL, raw notes, code snippet - Output: spoken script (~75–80 words) + optional B-roll/overlay production notes - No external tools required --- ## Workflow ### Step 1 — Parse Input & Infer Tone Extract from the user's input: 1. **Core message**: the single most interesting or useful thing to convey — state it in one sentence 2. **Audience**: infer from content — developers, students, general audience, creatives, etc. 3. **Tone**: pick from the table below based on content type and any user signals | Tone | When to use | Voice style | |------|-------------|-------------| | Tech/Educational | Code, tools, tutorials, "how to" | Calm, precise — "here's how it works" | | Casual/Discovery | Personal tips, life hacks, finds | Energetic — "okay so I just found out..." | | Dramatic | Big reveals, contrarian takes | Intense, punchy, short sentences | | Controversial | Unpopular opinions, hot takes | Bold, direct — "unpopular opinion:" | Default to **Tech/Educational** if context is ambiguous. Only ask the user if the content fits two very different tones equally. ### Step 2 — Write the Hook (3 seconds / 8–12 words) The hook must stop a viewer mid-scroll. Generate one hook per pattern, pick the strongest three: | Pattern | Template | Example | |---------|----------|---------| | Shocking stat | "[X]% of [audience] don't know [fact]." | "90% of devs don't know Python does this." | | Bold claim | "This [thing] [result] me [benefit]." | "This one decorator saved me 200 lines of code." | | Direct question | "Are you still doing [X] the hard way?" | "Are you still writing SQL joins by hand?" | | Pattern interrupt | "Stop. Nobody talks about [X]." | "Stop. Nobody talks about FastAPI's hidden feature." | Present all 3 hook variants to the user. Do not pick one for them — they know their audience best. ### Step 3 — Write the Core Content (20 seconds / 50–55 words) If the user hasn't specified, ask: *"Should the core summarize the topic, or ask the viewer a question to engage them?"* **Mode A — Summary:** 3 punchy sentences. Each sentence = one clear fact, step, or insight. **Mode B — Question/Engagement:** State the problem (1 sentence) → hint at the solution (1 sentence) → tease more detail, e.g. "and there's a catch I'll cover in part 2." Rules for both modes: - No sentence longer than 15 words - No jargon without a one-word definition immediately after it - Every sentence must deliver one discrete piece of value — cut filler words ruthlessly ### Step 4 — Write the CTA (5 seconds / 12–15 words) Use one of these templates. Do not write a generic "like and subscribe" — it converts near zero: | CTA type | Template | |----------|----------| | Subscribe ask | "Follow for one [topic] trick every day. You won't regret it." | | Comment engagement | "Drop a [emoji] in the comments if this saved you time." | | Question hook | "What's YOUR go-to trick for [topic]? Tell me below." | | Series tease | "Part 2 drops [day]. Follow so you don't miss it." | Pick the CTA that fits the tone. Content that naturally invites discussion → Comment or Question type. ### Step 5 — Assemble & Word Count Join: `[Chosen Hook] + [Core Content] + [CTA]` Count total words. Target: **75–80 words**. - Over 80 words: cut one sentence from core content, not the hook or CTA. - Under 70 words: add one concrete detail or example to the core. ### Step 6 — Production Notes (optional) Add 3–5 bullet suggestions for: - What to show on screen during the hook (face cam, code editor, product demo, screen recording) - Key phrases to display as on-screen text overlays (bold the phrase in the script to signal this) - Where to cut or transition --- ## Output Format ``` ## YouTube Short Script — [Topic] **Tone:** [Tech/Educational | Casual | Dramatic | Controversial] **Word Count:** [N] words (~[N] seconds) --- ### Hook Options (pick one before recording) 1. [Hook variant A] 2. [Hook variant B] 3. [Hook variant C] --- ### Full Script (using Hook 1) [Hook — 8–12 words] [Core content — 50–55 words, 3 sentences max] [CTA — 12–15 words] --- ### Production Notes - Hook visual: [what to show on screen] - Overlay: "[key phrase]" as on-screen text when mentioned - Cut at: [moment in the script] ``` --- ## Anti-Patterns - Never write a hook longer than 12 words — the viewer scrolls before the content starts. - Never write a generic CTA like "like and subscribe if you enjoyed this" — name a specific action. - Never exceed 80 words total — at normal speaking pace this runs past 30 seconds and gets cut off by the platform.
From aayushostwal/nexus, MIT.
More from aayushostwal/nexus
- 0.648score
software-engineer
aayushostwal
Resolve a GitHub issue or a concrete bug described in chat in a local repository, from reproduction through a verified fix and pull request.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
debugging
aayushostwal
Use when something is failing, regressing, or behaving unexpectedly and the goal is root-cause analysis, narrowest fix, and verification.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
nexus
aayushostwal
Use for persistent TODO capture, daily brief retrieval, and shared Codex/Claude operating rules.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
observability
aayushostwal
Use for correlated API failures, cascading errors, and distributed-system incident tracing.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
performance
aayushostwal
Use for memory, CPU, latency, and query-performance investigations, plus dependency-upgrade blast-radius analysis.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
reliability
aayushostwal
Use for production incidents, outages, and service degradations, plus release-readiness gate checks.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
skill-writer
aayushostwal
Use for creating, refactoring, or debugging skills and improving SKILL.md trigger quality.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
tutorial
aayushostwal
Use for creating executable Jupyter tutorials and AI engineering walkthroughs with runnable cells.
SkillAFreshness: A★ 19Claude Code +9 - 0.613score
testing
aayushostwal
Use for flaky or nondeterministic tests, especially local-vs-CI inconsistencies.
SkillAFreshness: A★ 19Claude Code +9 - 0.592score
token-optimizer
aayushostwal
Always-on token optimization rules.
SkillBFreshness: B★ 19Claude Code +9
Signals
- Stars
- 19
- Forks
- 0
- Open issues
- 2
- Last commit (this folder)
- Sep 30, 2026 (28 days ago)
- Last release
- Aug 10, 2026 (2 months ago)
From GitHub (fetched Oct 24, 2026).
Details
- License
- MIT
- Repository
- github.com/aayushostwal/nexus
- Website
- nexus-ai.aayushostwal.com
- First seen
- Oct 24, 2026
- Last checked
- Oct 24, 2026
- Source
- GitHubgithub-graphql
More from aayushostwal
- 0.648score
software-engineer
aayushostwal
Resolve a GitHub issue or a concrete bug described in chat in a local repository, from reproduction through a verified fix and pull request.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
debugging
aayushostwal
Use when something is failing, regressing, or behaving unexpectedly and the goal is root-cause analysis, narrowest fix, and verification.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
nexus
aayushostwal
Use for persistent TODO capture, daily brief retrieval, and shared Codex/Claude operating rules.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
observability
aayushostwal
Use for correlated API failures, cascading errors, and distributed-system incident tracing.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
performance
aayushostwal
Use for memory, CPU, latency, and query-performance investigations, plus dependency-upgrade blast-radius analysis.
SkillAFreshness: A★ 19Claude Code +9 - 0.643score
reliability
aayushostwal
Use for production incidents, outages, and service degradations, plus release-readiness gate checks.
SkillAFreshness: A★ 19Claude Code +9
Cite this page
shorts skill. Agent Mart. https://agentmart.dev/skills/aayushostwal-shorts (last updated 2026-10-24)