VOD Chaptering

    1

    Turns a stream or VOD transcript (SRT, VTT, Whisper, AssemblyAI JSON) into paste-ready YouTube chapters with a QC report that checks every timestamp. Python only, no network calls.

    Free

    1 useSecurity scanned

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    VOD Chaptering

    Example session with this skill installed

    I have an SRT transcript for my 2-hour Elden Ring stream. Can you generate some YouTube chapters for it? I just uploaded the file to the workspace.

    • Read your context and instructions
    • Compiled the vod chaptering

    I have analyzed the transcript for your Elden Ring VOD. I identified 7 distinct chapters based on your spoken transitions and gameplay activity changes. The chapters have been verified against the transcript timestamps and meet all YouTube formatting requirements.

    YouTube Chapters

    0:00 Starting soon
    3:15 Intro and build overview
    14:42 Exploring Limgrave
    42:10 Boss fight: Margit the Fell Omen
    1:05:30 Stormveil Castle entrance
    1:38:15 Finding the hidden basement
    1:55:04 Wrap up and superchats
    

    QC Report Summary

    • Status: PASS (0 FAIL, 2 WARN)
    • Warning [H1]: Found 7 chapters; suggested range for 2h VOD is 4-8. (Within range)
    • Warning [E1]: First quote for "Starting soon" starts 18s late due to silent intro screen.

    Next steps

    1. Copy the code block above and paste it into the top of your YouTube video description.
    2. Skim the section at

    0:00; it had no speech for the first three minutes, so check if the "Intro" starts exactly where you want it.
    3. Verify the title for the "hidden basement" segment, as the ASR had some difficulty with the specific room name.

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

    About this skill

    Manually chaptering hours of VOD transcripts is a slog that leads to inaccurate timestamps and broken YouTube links. This skill automates the analysis of transcripts to generate precise, verified YouTube chapters backed by word-for-word evidence.

    What it does

    • Multi-format parsing processes SRT, VTT, Whisper/WhisperX JSON, AssemblyAI, and YouTube json3 files.
    • Structural analysis identifies transitions based on speech patterns, activity changes, and no-speech gaps.
    • Automated verification runs a QC script to ensure every chapter meets YouTube's 0:00 start and 10s minimum rules.
    • Evidence matching pins each chapter to specific transcript quotes to prevent drift between speech and timestamps.
    • Activity detection flags "low confidence" sections like starting screens or silent gameplay for manual review.

    How it works

    1. Prep the transcript to generate a normalized speech digest and activity overview.
    2. Map the stream structure to identify topic shifts, game changes, or recurring segments.
    3. Pin boundaries using a window tool to find the exact line and timestamp where new content begins.
    4. Build the final output, running a validator that checks for title length, chapter count, and quote accuracy.

    Frameworks & tools

    This skill utilizes a specialized Python helper script for timestamp math and transcript normalization. It is compatible with any text-based or JSON transcript format, including those from OpenAI Whisper and major ASR services.

    Why this beats prompting it yourself

    Standard LLM prompting often results in estimated timestamps that drift or titles that hallucinate events not in the text. This skill uses a validation loop and strict evidence-based pinning to ensure the chapters actually line up with the audio.

    Use cases

    • Gaming VODs for marking specific levels, boss fights, or lobby waits.
    • Podcasts for separating interview segments, ad breaks, and Q&A.
    • VTuber streams for identifying intro screens, superchat readings, and game transitions.
    • Technical tutorials for breaking down long coding or build sessions into searchable steps.

    Known limitations

    Cannot see visual-only changes; if a game switch happens in total silence, it may be missed. Titles are capped at 40 characters to ensure visibility on mobile devices.

    This is the free sample of the VTuber Shorts Production System. It is not affiliated with YouTube, Twitch or any VTuber agency.

    Cannot detect visual-only changes during silence. Requires a timestamped transcript. Titles are limited to 40 characters for mobile compatibility.

    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 any payment.

    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

    1 use

    Times used to date

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 27 days ago

    • Free with an Agensi account

    Requires

    Python

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed27 days ago

    What's inside

    Frequently Asked Questions

    This skill uses a validation loop and strict evidence-based pinning to ensure every chapter lines up with the audio, preventing the timestamp drift and hallucinations common with standard LLM prompting.

    The skill supports a wide range of formats including SRT, VTT, Whisper/WhisperX JSON, AssemblyAI, and YouTube json3 files, making it compatible with almost any major transcription service.

    Yes, the skill includes an automated QC script that verifies every output meets YouTube's specific requirements, such as starting at 0:00 and maintaining a 10-second minimum chapter length.

    You receive the complete skill logic, a specialized Python helper script for timestamp normalization, and access to all future version updates as new transcription formats are supported.

    Since the skill relies on text analysis, it may miss transitions that occur in total silence (like a visual-only game switch); however, it flags these low-confidence sections for your manual review.

    Setup is straightforward: you simply provide your transcript file, and the skill’s internal Python tool handles the complex math and structural mapping automatically.

    New in Social Media