RonanCodes/youtube-to-pdf
skills/youtube-to-pdf/SKILL.md
Turn a YouTube talk into a beautiful, print-ready PDF "field guide" — transcript + scene-detected slide screenshots + a designed editorial layout with action points. Pulls the transcript and slides, understands the talk, then composes a multi-page A4 PDF using the proven field-guide template (cover, idea cards, stat callouts, pulled quotes, embedded slides, a grouped action-points checklist). Use when the user wants a PDF from a YouTube video, "youtube to pdf", a talk turned into a handout/guide/one-pager, or a designed summary of a video with screenshots. Sibling of /ro:video-summarize (extraction only) and the llm-wiki /generate pdf (vault pages -> pdf); this one is URL in, designed PDF out.
research
Updated Jun 30, 2026
$ install --global
skillsauthnpx skillsauth add RonanCodes/ronan-skills youtube-to-pdfInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
Security Scan Results
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
3
Scanners Passed
9
Scanners in report
Clean
TrivyContainer and dependency vulnerability scanner
95%
Clean
SemgrepStatic code analysis for vulnerabilities
95%
Clean
mcp-scan (Snyk)Model Context Protocol security validation
95%
Skipped
Snyk (dep)Open source security scanning
50%
Skipped
Socket.devSupply chain security analysis
50%
Skipped
VirusTotalMulti-engine malware detection
50%
Skipped
CrowdStrikeAdvanced threat intelligence
50%
Skipped
OSV-ScannerOpen Source Vulnerability database check
50%
Skipped
OWASP Dep-Check
50%
Last scanned: Jun 30, 2026, 2:29 AM140.1s3 files scanned
SKILL.md
- name:
- youtube-to-pdf
- description:
- Turn a YouTube talk into a beautiful, print-ready PDF "field guide" — transcript + scene-detected slide screenshots + a designed editorial layout with action points. Pulls the transcript and slides, understands the talk, then composes a multi-page A4 PDF using the proven field-guide template (cover, idea cards, stat callouts, pulled quotes, embedded slides, a grouped action-points checklist). Use when the user wants a PDF from a YouTube video, "youtube to pdf", a talk turned into a handout/guide/one-pager, or a designed summary of a video with screenshots. Sibling of /ro:video-summarize (extraction only) and the llm-wiki /generate pdf (vault pages -> pdf); this one is URL in, designed PDF out.
- category:
- research
- argument-hint:
- <youtube-url> [--out <path.pdf>] [--title "..."] [--ingest <vault>]
- allowed-tools:
- Bash, Read, Write, Edit, WebFetch
YouTube to PDF
Take a YouTube talk and produce a PDF a person actually wants to read: the real slides as screenshots, the load-bearing ideas, the quotes, and a concrete action-points checklist, laid out like an editorial field guide rather than a wall of text. This is the pipeline that produced the Farhan Thawar "What Is Your Job Now" guide.
The output is beautiful by contract, not by luck. The design lives in assets/example-field-guide.html (a complete, proven, print-correct document). You keep its <head>/<style> verbatim and rewrite the page bodies for the new talk.
Pipeline
1. Extract
bash scripts/extract-youtube.sh "<url>" <work-dir>
Produces in <work-dir>: meta.txt (title, uploader, duration, chapter markers), transcript-clean.md (timestamped), frames/ (scene-detected slides), frames/contact-sheet.png (a montage), frames/order.txt.
2. Understand the talk
READ transcript-clean.md in full. This is the step that makes the PDF good: you must actually understand the argument, not skim it. Note the thesis in one line, the 5-9 load-bearing ideas, the best verbatim quotes (with timestamps), the concrete numbers/examples, and a real set of action points (imperative, specific, grouped by who acts).
3. Pick the slides
VIEW frames/contact-sheet.png (Read it as an image). Scene detection catches camera cuts as well as slides, so pick only the genuine content slides (the clean full-screen ones), 8-14 of them. Copy those out of frames/ into a slides/ folder next to your HTML, renamed descriptively (e.g. 02-sdlc-shift.png). Resize wide if huge: convert in.png -resize 1280x out.png. Slides are the soul of this format; do not skip them.
4. Compose the HTML
Copy assets/example-field-guide.html to <work-dir>/index.html. Keep the entire <head> and <style> block verbatim (it is print-tuned: A4 page size, fixed-height pages, the palette, the components). Replace the <section class="page"> bodies with your content using the component palette below. Reference slides as slides/<name>.png (relative to the HTML).
Aim for 5-8 pages: cover, the core argument (with a slide + a diagram), the key ideas as cards, any "operating manual" / numbered list, a deep-dive on the single best idea (with its slide), the action-points checklist, and a closing/bridge page. Set a real <title>.
5. Render
bash scripts/render-pdf.sh <work-dir>/index.html <out.pdf>
Prints the page count.
6. Verify it fits (REQUIRED — this is the no-overflow gate)
READ the rendered PDF pages back as images. Check EVERY page: no text runs under the footer, no blank trailing page, no clipped figure. Each <section class="page"> is one fixed A4 sheet with overflow:hidden, so content that is too tall silently collides with the footer (the most common defect). If a page overflows: cap that figure (style="max-height:70mm; width:auto; margin:0 auto;"), shorten copy, or move a block to a new page. Re-render and re-check until clean.
7. Open it
open <out.pdf> (macOS). If --ingest <vault> was passed, also drop the PDF in that vault's artifacts/pdf/ and the transcript in raw/ (or hand off to the llm-wiki /ingest skill for full wiki pages).
The design contract (what makes it beautiful)
- A4, fixed pages.
@page{size:A4;margin:0}and.page{height:297mm; overflow:hidden}. Never let a page reflow past its sheet. - Real slides, embedded. Screenshots carry the talk. Caption each one in italics.
- Editorial, not a report. Serif display headings, generous whitespace, an amber accent rule under each section title, pulled quotes between hairlines, stat callouts with one huge number.
- Action points are the payoff. A grouped checklist (by who acts), each item bold-lead + a faint "why". This is usually what the reader keeps.
- Plain copy. No em-dashes or en-dashes; no AI-tell words (delve, leverage, robust, seamless, elevate, unlock, streamline). Specific over abstract: use the real numbers and names from the talk.
Component palette (copy from assets/example-field-guide.html)
- Cover
section.page.cover: kicker + big serifh1(accent one word) + italic.sub+ a.thesisblockquote + a.metarow. - Section page:
.kicker+h2.sec+.sec-rule+.leadparagraph. - Idea cards
.grid2 > .card: numbered.n,h4, one-linep. Great for "N things to know". - Stat callout
.stat: a huge.bignumber + an explanatory.txt. Use for the killer figures. - Pulled quote
.quote+.who: a verbatim line with attribution + timestamp. - Figure
figure > img + figcaption: an embedded slide with an italic caption. - Numbered manual
ol.big > li > h4 + p: for an ordered "how we work" list. - Action checklist
.actionswith.grpgroup headers and.actrows (each a.box+ bold lead +.why). - Bridge table
.tbl: a clean 3-column table for a mapping/summary. - Footer
.foot: left = doc name, right = section name. One per page.
Notes
- Extraction reuses the same tools as
/ro:video-summarize; if that skill is installed you may call it instead ofscripts/extract-youtube.sh, then continue from step 2. - For a PDF built from wiki pages rather than a video, use the llm-wiki
/generate pdfskill (it now shares this field-guide look). - Keep the template's
<style>in sync: if you improve the design, updateassets/example-field-guide.htmlso every future guide inherits it.
Related Skills
RonanCodes/skills/linear-pipeline
testing
VerifiedTrustedCommunity
--- name: linear-pipeline description: The Fable orchestrator for a single dispatched Linear ticket. Holds almost no context itself; it receives `--issue <ID> --detached`, decides the stage sequence, and fans out a sub-agent per stage, passing forward only each stage's artifact (never re-derived, never inlined into its own context). Step zero, before any planning or stage routing, is a boundary triage against `canon/security-boundary.md` (#199): a match tags Ronan Connolly and stops the run, no
SKILL.mdUpdated Jul 5, 2026
RonanCodes/skills/in-your-face
development
VerifiedTrustedCommunity
--- name: in-your-face description: Capture a chat-only answer into a durable artifact (markdown + HTML, PDF when cheap) and launch it automatically so the user cannot miss it. Use when user says "in your face", "don't let me lose this", "save that answer", "make that durable", or right after answering a substantive side question (a recipe, comparison, how-to, or generated prompt) that would otherwise die with the context. category: workflow argument-hint: [--no-open] [--vault <short>] [hint of
SKILL.mdUpdated Jul 5, 2026
RonanCodes/codex
tools
VerifiedTrustedCommunity
One-shot headless OpenAI Codex CLI calls for background/admin AI tasks — summaries, classification, extraction, admin glue. The default engine for anything that runs AI constantly in the background (daemon-driven, per-event), because it bills the flat ChatGPT subscription instead of Claude usage or per-token API spend, and it keeps working while Claude is rate-limited. NEVER for coding — coding stays Claude. Use when a skill or daemon needs a cheap always-on AI call, when the user says "use codex", "ask codex", "codex as backup", or when building a background summarizer/classifier into a listener or loop. Reads auth from ~/.codex/auth.json (ChatGPT account, no API key).
SKILL.mdUpdated Jul 5, 2026
RonanCodes/warranty-check
research
VerifiedTrustedCommunity
Turn a warranty rejection, repair quote, or RMA email into a cited decision brief — legal read (NL/EU consumer law), is the part user-serviceable, live part and new-unit prices, repair-vs-DIY-vs-new economics, before-you-send-it checklist, deadlines. Use when the user pastes or screenshots a repair quote, warranty rejection, "not covered" email, onderzoekskosten fee, or asks "should I repair or replace this".
SKILL.mdUpdated Jul 4, 2026