build-in-public-youtube
Use when Dhruv shares a long-form (1-3 hour) team meeting recording or transcript and wants to publish it on YouTube as build-in-public content. Produces title, description, thumbnail prompt, tags, chapters, and pinned comment. Works from RTF, TXT, MD transcripts or audio files (with transcribe.py).
aiagentwithdhruv · 已有 AI 审核记录
安装与使用
运行命令前先阅读来源与依赖说明。复制命令不会自动安装或执行。
npx skills add aiagentwithdhruv/skills --skill build-in-public-youtube技能说明
name: build-in-public-youtube description: Use when Dhruv shares a long-form (1-3 hour) team meeting recording or transcript and wants to publish it on YouTube as build-in-public content. Produces title, description, thumbnail prompt, tags, chapters, and pinned comment. Works from RTF, TXT, MD transcripts or audio files (with transcribe.py).
Build-in-Public YouTube — Long-Form Team Meeting → YouTube Episode
What this skill does
Turns Dhruv's long Saturday/Sunday team sync recordings (Afeefa, Shailendra, Vishal, Navpreet + Dhruv) into a complete YouTube upload pack:
- Title (clickable, honest, build-in-public framed)
- Description (chapter timestamps, what's covered, links, team credits, hashtags)
- Thumbnail prompt (Gemini-paste-ready, brand palette: deep dark + neon cyan #18D1E0 + metallic grey)
- Tags (25-30 YouTube tags)
- Chapters (with timestamps if known, otherwise structural placeholders)
- Pinned comment (CTA + community + product links)
This is the canonical format for the AIwithDhruv "Build-in-Public Diaries" series.
When to invoke
Trigger phrases from Dhruv:
- "Here's the Sunday team recording, write a YouTube post"
- "Make a YouTube title for this transcript"
- "We did 2 hours yesterday, can we upload it"
- "Build-in-public episode from this"
- Or just shares a
.rtf/.txt/.mdtranscript insideLibrary/Mobile Documents/com~apple~TextEdit/Documents/orinbox/recordings/
Series identity (locked)
| Field | Value |
|---|---|
| Channel | AIwithDhruv (main) |
| Series name | Build-in-Public Diaries |
| Episode prefix | Build-in-Public Diaries — Ep N |
| Cadence | 1-2 long-form episodes per week (weekend syncs + Wednesday mid-week sync if any) |
| Duration | 1-3 hours unedited (NO heavy editing — raw is the brand) |
| Voice | First-person from Dhruv ("we built", "I shipped", "we lost", never "the team did") |
| Brand palette | Deep dark #0A0E1A + neon cyan #18D1E0 + metallic grey. NO orange / red / teal-mint. (See brand_palette.md) |
| Format | 16:9, 1080p min, thumbnails 1280x720 |
Inputs you need before generating
If any of these are missing, ASK Dhruv before drafting:
- Episode number (default to next sequential: check the channel or ask)
- Recording date (e.g., "Apr 26, 2026") — used in title + filename
- Recording length in HH:MM (estimate from word count if no audio: ~150 words/min)
- Path to transcript (
.rtf/.txt/.md) - Public revenue claim allowed? (default: NO. Don't drop ₹/$ numbers without explicit approval)
- Specific milestone to lead with? (e.g., "Vishal joined", "AutoBOQ shipped", "first paid client")
If you have the transcript path but no metadata, run wc -w <file> to estimate length and propose Episode N + date + suggested headline angle. Then wait for confirmation.
Step-by-step process
Step 1 — Read the transcript fully
Convert RTF if needed: textutil -convert txt -stdout "input.rtf" > /tmp/transcript.txt
Read end-to-end. Don't skim. Look for:
- Strategic decisions (pivots, hires, product calls)
- Real numbers (only use them if Dhruv approved public-revenue mode)
- Tools/stack mentioned (Claude, Antigravity IDE, Blotato, Plivo, Flow, Vercel, etc.)
- Honest losses / blockers (these make the best build-in-public hooks)
- Personal moments (team onboarding, philosophy, framework calls)
Step 2 — Extract the 5 anchor points
Every episode pack is built on these 5 anchor points pulled from the transcript:
- The headline decision — the one thing this meeting will be remembered for
- The week's score — wins shipped, losses absorbed
- The team motion — who did what, who joined, who graduated
- The technical reveal — one stack/tool/pattern that was tested or chosen
- The next-7-days commit — public commitment so viewers can hold us accountable
Write these 5 lines first. Title and description flow from them.
Step 3 — Generate the upload pack
Apply the templates in Reference Templates below. Always show the full pack to Dhruv for approval before suggesting upload. Never auto-publish (see feedback_never_auto_publish.md).
Step 4 — Save the pack
Write the final pack to /tmp/<recording-filename>_episode_pack.md so Dhruv can copy-paste into YouTube directly. Print location at end.
Reference Templates
Title formulas (pick one, A/B is fine)
Formula 0 — Setup / Foundation video (use ONCE, for the genesis "operating manual" video):
{$0 MRR / honest current state}, {$tool spend}, {N products}: How {brand} Actually Works ({Setup / Operating Manual / Day 0})
Example: $0 MRR, $400/mo Tools, 4 Products: How AIwithDhruv Actually Works (Build-in-Public Setup)
Use this only for the first/foundation video that explains how the team and stack are wired. NOT for ongoing weekly episodes.
Formula A — Decision lead (best for big pivots/hires, ongoing episodes):
Build-in-Public Diaries — Ep {N}: {THE ONE DECISION} | Day {X} of the 45-Day Sprint
Example: Build-in-Public Diaries — Ep 3: We Hired 4 People in One Weekend | Day 12 of the 45-Day Sprint
Formula B — Stakes lead (best for weeks with money/risk on the line):
{X Hours} of Real Strategy: How We're Trying to Hit ₹{Y} in {Z} Days (Build-in-Public Diaries Ep {N})
Use only if revenue mode = approved.
Formula C — Team lead (best when a meeting introduces team):
Meet the Team Building 4 AI Products in Public — Sunday Sync Ep {N}
Formula D — Stack lead (best when one tool/tech is the star):
We Replaced {Old Tool} With {New Tool} — Live Decision in Sunday Sync (Build-in-Public Ep {N})
Formula E — Honesty hook (use anytime real numbers are public):
{$honest current state}: Here's the {N-Day} Plan to Change That ({Build-in-Public Ep N})
Example: `We're at $0 MRR. Here's the 45-Day Plan to Land Our First Client (Build-in-Public Ep
阅读完整来源 ↗