More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    video study notes

    1

    Turning lecture videos and podcasts into usable revision material often requires hours of manual pausing and typing.

    $8.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    video study notes

    video study notes

    Example session with this skill installed

    Take notes from this lecture and turn it into something I can actually revise from: timestamped notes, a one-page cheat sheet, practice questions and a study plan. The video is in Chinese — write everything in English.

    https://www.youtube.com/watch?v=zjyGtQ4uKwg

    • Read your context and instructions
    • Compiled the video study notes
    • Generated the data export

    What it did

    Fetched the lecture's Chinese captions (18:48, 48 segments), processed them in two chunks, merged them into one note set, validated it, and rendered the pack. It ran on the agent's own model — no second API key.

    FileSizeWhat it is
    notes.md13 KB4 chapters, 15 sections, exam points, glossary
    cheatsheet.md2.5 KB6 formulas + 7 definitions, one printable page
    quiz.md4 KB9 questions, 3 tiers, full answer key
    anki.csv5.5 KB20 import-ready cards
    review-plan.md1.6 KB7 dated sessions, D0 → D30
    index.html34 KBone offline page, clickable timestamps

    Validated, not estimated

    VALID: 0 error(s), 0 warning(s)
    chapters 4, sections 15
    

    Every timestamp is checked against a real caption segment start, so "00:14:47" really is 00:14:47. A pack that fails the check is not rendered.

    The notes (excerpt)

    4. Policy › 4.3 Policy as a conditional probability 00:14:47

    For anything beyond a picture, the policy is written as π: the probability of an action given the state. For a deterministic policy on s1, moving up has probability 0, moving right 1, and every other action 0 — and the probabilities always sum to one:

    π(a | s),   Σ over a in A(s) of π(a | s) = 1
    

    A stochastic policy is not a broken deterministic one; the randomness is the policy.

    The questions (excerpt)

    q4 · medium · single choice · 00:06:50

    From s1 the agent takes a1 (up) and stays in s1. Why?

    • A. The transition is probabilistic
    • B. The boundary behaves like a wall and bounces the agent back
    • C. a1 is not a valid action in s1
    • D. The agent chose to stay

    Answer B. The boundary blocks the move, so the next state is s1 again. This is a modelling choice: in a game the bounce could have been defined to land in s4 or s7.

    One practice run, and the plan reacts

    SCORE 5.00 / 9  (56%)
      basic 2.00/3   medium 2.00/3   advanced 1.00/3
    MISSED
      q1 (basic, judge) expected false, got true — Action space
      q5 (medium, single) expected B, got A — Transition probability
    WEAK POINTS Action space, Transition probability
    

    Running the build again with --review review.json re-renders the plan with those two points scheduled into the next sessions.

    Next steps

    Open index.html, tick each chapter as you read it, take the quiz before session 1, and print cheatsheet.md the night before the exam.

    anki.csv

    CSV · data export

    Generated

    Example file from a real run - the skill writes it into your workspace.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Turning lecture videos and podcasts into usable revision material often requires hours of manual pausing and typing. This skill automates the pipeline from transcript to structured study pack. It processes captions from platforms like YouTube, Bilibili, and local media to generate timestamped notes, tiered quizzes, and spaced-repetition schedules without requiring external API keys.

    What it does

    • Structured Note Generation creates timestamped explanations, derivations, and formulas instead of simple transcript summaries.
    • Tiered Assessment builds question banks with multiple-choice, calculation, and short-answer formats across basic, medium, and advanced difficulty levels.
    • Flashcard Export generates CSV files formatted for direct import into Anki.
    • Automated Scheduling produces an Ebbinghaus spaced-repetition plan (D0 to D30) based on content or quiz performance.
    • Interactive Study Page renders a single-file, offline HTML dashboard with searchable content and clickable timestamps that jump to the video.

    How it works

    1. Fetch Transcript using provided Python scripts to extract captions and metadata from a URL or local file into JSON chunks.
    2. Map & Reduce the transcript chunks into a validated JSON source of truth, organizing content by subject-specific styles like math or law.
    3. Validate & Build the pack using automated check scripts to ensure every timestamp and formula is accurate and anchored to the source.
    4. Render the final outputs including the printable cheat sheet, Anki deck, and interactive index page.

    Frameworks & tools

    This tool uses yt-dlp for caption retrieval and faster-whisper for local transcription. It operates via Python scripts that generate Markdown, JSON, and LaTeX.

    Why this beats prompting it yourself

    Generic prompts often invent timestamps or miss critical technical formulas. This skill uses a multi-step validation loop and strict JSON schemas to ensure every claim is anchored to a specific second in the video.

    Use cases

    • Convert a technical YouTube playlist into a structured course with a unified revision plan.
    • Generate an exam-cram sheet and 20-question practice test from a recorded university lecture.
    • Build a flashcard deck from a podcast for long-term retention of key concepts.
    • Create an offline, searchable archive of training videos with interactive timestamps.

    Known limitations

    Requires videos to have captions or requires local installation of faster-whisper for audio processing. Does not support video editing, downloading media files, or translating subtitles.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Education & Learning.

    Requires captions or local faster-whisper for audio. No video/audio downloading. Timestamps must exist in source transcript. Python and yt-dlp dependencies required for fetching.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 17 days ago

    Software
    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Network Access
    Listed17 days ago

    What's inside

    Frequently Asked Questions

    This skill generates structured, timestamped notes, a tiered quiz bank, a CSV file for Anki flashcards, an Ebbinghaus-based study schedule, and an interactive offline HTML dashboard.

    Unlike generic prompts that can hallucinate details, this skill uses a multi-step validation loop and strict JSON schemas to ensure every formula and timestamp is anchored directly to a specific second in the source video.

    The skill processes URLs from platforms like YouTube and Bilibili using yt-dlp, as well as local media files; if a video lacks captions, it can perform local transcription using the integrated faster-whisper tool.

    You receive the core Python processing scripts, the validation logic, and the rendering templates needed to transform raw transcripts into the final study dashboard and flashcard exports.

    This skill is designed for local execution via Python, ensuring your study data stays private and eliminating the need for recurring external API costs for transcription or processing.

    The output is a single-file, offline HTML dashboard that allows you to search content and click timestamps to jump directly to the relevant part of the video in your browser.

    New in Education & Learning