youtube-to-markdown

A workflow for turning YouTube videos into structured Markdown files. YouTube is a video-sharing website, and a transcript is the written version of spoken video content.

In plain words
What is it for?
Use it to collect video details, create a cleaned transcript, summarize the video, and analyze comments.
Why use it?
It combines video extraction, transcript cleanup, summaries, and comment analysis into a defined process.

Skill for Claude CodeCodex

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add skills/vre/flow-state/combined
Any agent
npx skills add vre/flow-state --skill combined
Clone the repo
git clone --depth 1 https://github.com/vre/flow-state

Made for: Claude Code, Codex.

Per session 35 Skills are progressive disclosure: only the name and description are preloaded; the body loads when the skill is used.
When invoked 1,238 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00035 $0.01238
Opus 5 $0.00017 $0.00619
Sonnet 5 $0.00007 $0.00248
Haiku 4.5 $0.00003 $0.00124

Measured 2d ago against content hash 57185d84aaa9, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

youtube-to-markdown scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

docs/youtube-to-markdown/research/2026.01.14 token usage test/code/combined/SKILL.md · 184 lines

How it starts

The opening of the file, as written. The whole thing — 184 lines — stays where its author put it; the contents beside it link to each section on GitHub.

YouTube to Markdown (Combined Variant)

Test variant: Keeps all polish steps (4, 7, 8). Combines summary (5+6) and comments (10a+10b) into single subagents.

Execute all steps sequentially without asking for user approval.

Step 0: Check extracted before

python3 ./check_existing.py "<YOUTUBE_URL>" "<output_directory>"

If returns exists: true: Skip to Step 10 (comments).

Step 1: Extract data

python3 extract_data.py "<YOUTUBE_URL>" "<output_directory>"

Step 2: Extract transcript

If video language is en, proceed directly. If non-English, ask user which language.

python3 extract_transcript.py "<YOUTUBE_URL>" "<output_directory>" "<LANG_CODE>"

Fallback (only if transcript unavailable)

python3 extract_transcript_whisper.py "<YOUTUBE_URL>" "<output_directory>"

Step 3: Deduplicate transcript

python3 ./deduplicate_vtt.py "<output_directory>/${BASE_NAME}_transcript.vtt" "<output_directory>/${BASE_NAME}_transcript_dedup.md" "<output_directory>/${BASE_NAME}_transcript_no_timestamps.txt"

Step 4: Add natural paragraph breaks

task_tool:

  • subagent_type: "general-purpose"
  • model: "sonnet"
  • prompt:
INPUT: <output_directory>/${BASE_NAME}_transcript_no_timestamps.txt
CHAPTERS: <output_directory>/${BASE_NAME}_chapters.json
OUTPUT: <output_directory>/${BASE_NAME}_transcript_paragraphs.txt

Analyze INPUT and identify natural paragraph break line numbers.
Read CHAPTERS. If contains chapters, use as primary break points.
Target ~500 chars per paragraph.

Write to OUTPUT: 15,42,78,103,...
python3 ./apply_paragraph_breaks.py "<output_directory>/${BASE_NAME}_transcript_dedup.md" "<output_directory>/${BASE_NAME}_transcript_paragraphs.txt" "<output_directory>/${BASE_NAME}_transcript_paragraphs.md"

Step 5: Summarize transcript (combined summarize + tighten)

task_tool:

  • subagent_type: "general-purpose"
  • model: "sonnet"
  • prompt:
INPUT: <output_directory>/${BASE_NAME}_transcript_no_timestamps.txt
OUTPUT: <output_directory>/${BASE_NAME}_summary_tight.md
FORMATS: ./summary_formats.md

1. Classify content type:
   - TIPS: gear reviews, rankings, practical advice
   - INTERVIEW: podcasts, conversations, Q&A
   - EDUCATIONAL: concept explanations, analysis
   - TUTORIAL: step-by-step instructions

2. Read FORMATS and create summary using format for detected type.

3. Self-review: Cut fluff, enforce <10% of transcript bytes, prefer lists over prose.

4. Save final tightened summary to OUTPUT.

Rules:
- Skip ads, sponsors, self-promotion
- Preserve original language

ACTION REQUIRED: Use Write tool NOW to save to OUTPUT file.

Read the full file on GitHub · 184 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 184 lines · 35 tokens per session scan A 57185d84aaa9

Subscribe to this mod's changes

youtube-to-markdown is a skill published in the GitHub repository vre/flow-state (12 stars, last pushed 2mo ago), licensed MIT. It adds 35 tokens to every session and 1,238 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

Related

Other skills, from other repositories

ask-zilliz

Zilliz Cloud onboarding and usage assistant. Helps users understand Zilliz Cloud, choose the right plan, estimate costs, write code, debug issues, and adopt new features like Functions, Volumes, and Global Clusters. Use this skill whenever the user asks about Zilliz Cloud — including plan selection, pricing, cost…

zilliztech/zilliz-plugin · 236 tokens

diagnose

Use when the user reports that a Zilliz Cloud cluster or Milvus collection is unhealthy, slow, stuck, returning errors, hitting quotas, or otherwise misbehaving — or when they ask "what's wrong with...", "why is ... slow", "diagnose ...", "troubleshoot ...".

zilliztech/zilliz-plugin · 65 tokens

collection

Use when the user wants to create, list, describe, drop, rename, load, release, or manage collections and collection aliases in Milvus.

zilliztech/zilliz-plugin · 32 tokens

cluster

Use when the user wants to create, list, describe, delete, suspend, resume, or modify Zilliz Cloud clusters.

zilliztech/zilliz-plugin · 28 tokens

external-collection

Use when the user wants to trigger, describe, or list refresh jobs for an external collection (a collection backed by an external data source such as Vector Lake). Note this is the data-plane refresh workflow, not collection CRUD -- for create/drop/load see the collection skill.

zilliztech/zilliz-plugin · 58 tokens

backup

Use when the user wants to create, list, describe, delete, export, or restore backups, or manage backup policies on Zilliz Cloud.

zilliztech/zilliz-plugin · 32 tokens