Applied AI-Powered Music Production Workflows

The RECORDED LOCATIONS Episode Builder

Last reviewed

Beginner

What you'll learn

~20 min
  • Build a folder-to-publish CLI workflow for a YouTube series
  • Generate episode titles, descriptions, timestamps, and thumbnail concepts
  • Create a consistent publishing checklist for every episode
  • Integrate series branding and tracklist formatting into automated output

What you’re building

RECORDED LOCATIONS is your YouTube series — you record sets in different locations, each episode has its own vibe, tracklist, and story. But turning raw footage into a published YouTube video means a pile of metadata work: writing a description, formatting the tracklist with timestamps, coming up with title options, drafting thumbnail text, and going through a publish checklist so you don’t forget to add end screens or set the premiere time.

You’re going to build a tool that takes a folder of episode notes and generates all of that in one command.


The episode folder

For each episode, create a folder with a simple notes file. The folder structure looks like this:

episodes/s2e04-warehouse/
notes.json

Here’s what notes.json looks like (your agent can create this sample file for you when you start the build):

REFERENCE DATA episodes/s2e04-warehouse/notes.json sample episode metadata open reference hide reference Reference example of the episode notes JSON file your builder reads. Your agent can create or update this file for you from your rough tracklist notes.
{
"series": "RECORDED LOCATIONS",
"season": 2,
"episode": 4,
"location": "The Warehouse, Richmond VA",
"date": "2026-02-22",
"mood": "Dark, industrial, raw energy",
"duration": "58:30",
"tracklist": [
{ "time": "0:00", "artist": "moodmixformat", "title": "JADED", "label": "Unreleased" },
{ "time": "6:15", "artist": "Anyma", "title": "Consciousness", "label": "Afterlife" },
{ "time": "12:40", "artist": "moodmixformat", "title": "KICK IT W/U", "label": "Self-released" },
{ "time": "19:05", "artist": "CamelPhat & Anyma", "title": "The Sign", "label": "Afterlife" },
{ "time": "25:30", "artist": "moodmixformat", "title": "LIVE IT UP", "label": "Self-released" },
{ "time": "32:00", "artist": "moodmixformat", "title": "Untitled ID 2", "label": "Unreleased" },
{ "time": "38:45", "artist": "moodmixformat", "title": "Untitled ID 3", "label": "Unreleased" },
{ "time": "45:10", "artist": "moodmixformat", "title": "MOOD", "label": "Self-released" },
{ "time": "51:30", "artist": "moodmixformat", "title": "Untitled ID", "label": "Unreleased" }
],
"highlights": [
"First time playing JADED live",
"Crowd reaction during the KICK IT W/U drop",
"Surprise B2B with a local DJ for the last 10 minutes"
],
"gear": "Pioneer CDJ-3000 x2, DJM-V10, RMX-1000"
}
💡Keep the notes simple

You don’t need perfect timestamps or detailed descriptions. Even rough notes work — the tool formats and polishes everything. The key is getting the tracklist order and approximate timestamps right. You can always fix timestamps after watching the final edit.


Before you start

First, if your AI tool is still open, leave it: type /exit and press Enter (if you use Antigravity, press Ctrl+D with nothing typed instead), and wait until you see the $ or % prompt. Make a clean folder for this lesson inside ~/producer-workbench:

Terminal window
mkdir -p ~/producer-workbench/03-recorded-locations-builder && cd ~/producer-workbench/03-recorded-locations-builder

On Windows PowerShell, run mkdir -Force ~/producer-workbench/03-recorded-locations-builder; cd ~/producer-workbench/03-recorded-locations-builder, or open the folder in your desktop agent app. Save episodes/s2e04-warehouse/notes.json inside it (or ask your agent to create the sample file from the JSON above), then start your agent the way your track’s setup lesson showed.

Switch to plan mode

Once your agent is open, switch to Plan mode inside the session (Shift+Tab, /plan, or your desktop app’s mode picker). In plan mode it reads your episode notes, asks clarifying questions, and shows you a plan before building.

The conversation

Each message below is an example. Say it in your own words.

Say what you want

Start with a short plain-English description of what you want to build:

I want to build a tool that reads an episode folder's notes.json and generates YouTube titles, a timestamped description, thumbnail concepts, a publishing checklist, and chapters.json. Ask me what you need first.

Answer its questions

Your agent will ask a few questions about output locations, formatting, and YouTube rules. Answer each briefly:

If it asksYou might say
Where should the generated files be saved, and what are their names?Write titles.md, description.md, thumbnail.md, checklist.md, and chapters.json inside the same episode folder as notes.json.
What title variations and character limits should titles.md follow?Five options under 100 characters: standard series format, mood-based, tracklist teaser, location-focused, and a punchy Shorts title.
How should the YouTube description and tracklist timestamps be formatted?Short intro, then one track per line as “0:00 Artist - Title [Label]”, plus gear, links, and tags.
What should thumbnail.md and checklist.md include?Three visual thumbnail concepts with 4-word text overlays, and a markdown checkbox list covering pre-upload, upload, and post-publish steps.
How should chapters.json handle the tracklist and missing data?List each chapter’s time and title starting at 0:00, and write “Tracklist TBD” if the tracklist is empty.

Read the plan

Before approving, check the plan for four domain essentials:

  1. Output files land inside the episode folder. Running node episode-builder.js episodes/s2e04-warehouse/ writes all five files next to notes.json, not in the current working directory.
  2. YouTube chapter rules are enforced. The first timestamp in both description.md and chapters.json starts at 0:00, timestamps stay in ascending M:SS or H:MM:SS order from notes.json, and each track line follows 0:00 Artist - Title [Label].
  3. Titles stay under YouTube’s 100-character limit. All five title variations in titles.md stay concise and scannable, and the description intro avoids repeating the full tracklist in prose.
  4. Empty tracklists do not crash the CLI. If an episode folder has a notes.json without a tracklist yet, the tool writes "Tracklist TBD" in the description and skips chapters.json cleanly.

If the first plan writes files to the root folder or invents new timestamps instead of reading notes.json, say so briefly:

Write all output files inside the episode folder passed on the command line, and use the exact timestamps from notes.json starting at 0:00.

When the plan covers all four, approve it: pick the option to proceed or say Yes, go ahead and build it (in Codex, press Shift+Tab first to leave plan mode). Once you verify the five files appear in episodes/s2e04-warehouse/, ask your agent to save a Git checkpoint:

Save a checkpoint with Git so we can roll back if needed.

What you get

Ask your agent to run the episode builder on your sample folder and show you the generated files:

Run the episode builder on episodes/s2e04-warehouse/ and show me the generated files.

If you prefer running it in a terminal yourself, run node episode-builder.js episodes/s2e04-warehouse/. Your episode folder now looks like this:

episodes/s2e04-warehouse/
notes.json (your input)
titles.md (5 title options)
description.md (ready-to-paste YouTube description)
thumbnail.md (3 thumbnail concepts)
checklist.md (publishing checklist)
chapters.json (YouTube chapters data)

Verify each file

titles.md — Are the titles under 100 characters (YouTube’s title limit)? Do they capture the vibe? Pick one or combine elements from multiple options.

description.md — This is the big one. Copy-paste it into YouTube’s description field. Check:

  • Are timestamps formatted correctly? (YouTube turns description timestamps into chapters when the first is 0:00 and there are at least three, in ascending order)
  • Is the tracklist in the right order?
  • Do the social links have placeholders you can fill in?

thumbnail.md — These are concepts, not finished thumbnails. But they give you direction for when you open Photoshop or Canva.

checklist.md — Go through it item by item. Does it cover everything you usually forget?

chapters.json — This can be used programmatically if you ever build a more automated upload pipeline.

ℹThe tracklist is your secret weapon

A properly formatted tracklist with timestamps is one of the most viewer-friendly things you can add to a YouTube description. It turns a 58-minute set into navigable chapters. Viewers come back to specific tracks, which boosts watch time and engagement. The tool formats this perfectly every time.


If something is off

ProblemFollow-up prompt
Timestamps don’t match tracklistThe timestamps in the description don't match the tracklist from my notes.json. Use the exact timestamps from the tracklist in notes.json, don't generate new ones.
Titles are too longSeveral titles are over 100 characters, which is YouTube's limit. Shorten all options to under 80 characters while keeping them descriptive.
Description feels like a blog postThe description opening is too long and reads like a blog post. Make it 2 sentences max. The audience clicked on a DJ set video, they don't need 3 paragraphs of context.
Checklist is too basicExpand the checklist with export resolution (1080p60 vs 4K), a consistent audio loudness target (like -14 LUFS), premiere vs instant publish, end screen timing in the last 5-20 seconds, and adding the video to the RECORDED LOCATIONS playlist.

Customize it

Auto-generate SoundCloud description

Add a soundcloud-description.md file with a plain artist-and-title tracklist without timestamps, a link to the YouTube video, and SoundCloud tags.

Add a social media snippet pack

Add a social.md file with copy-paste blocks for a 2-3 word Instagram story overlay, a 3-4 line feed caption, and a punchy X post.

Template for future episodes

Create a template.json file with placeholder fields for new episodes, plus a README.md explaining the episode folder structure and how to run episode-builder.

🔧

When Things Go Wrong

Use the Symptom → Evidence → Request pattern: describe what you see, paste the error, then ask for a fix.

Symptom
The tool creates files in the wrong directory
Evidence
Running `node episode-builder.js episodes/s2e04-warehouse/` creates the output files in the current directory instead of inside the episode folder
What to ask the AI
"The output files are being written to the current directory instead of the episode folder. Write all output files to the same directory that contains the notes.json file. Use the input path to determine the output directory."
Symptom
The tracklist has duplicate entries
Evidence
The description.md shows each track listed twice -- once in the opening paragraph and once in the tracklist section
What to ask the AI
"The tracklist is duplicated in the description. The opening paragraph should reference the mood and location, not list individual tracks. Keep the full tracklist only in the dedicated timestamp section."
Symptom
The chapters.json has incorrect time format
Evidence
YouTube chapters need timestamps like '0:00' but the JSON has values like '0' or '0:0:00'
What to ask the AI
"Make the chapter timestamps consistent: the first must be '0:00', use 'M:SS' for times under an hour (e.g., '6:15', '38:45') and 'H:MM:SS' after that. Every value needs minutes and seconds, never a bare '0'."
Symptom
The tool crashes when tracklist is empty
Evidence
I have an episode with no tracklist yet (just location and mood) and the tool throws 'Cannot read property of undefined'
What to ask the AI
"The tool crashes when the tracklist array is empty or missing. Handle the case where tracklist is empty: skip the tracklist section in the description, note 'Tracklist TBD' in the description, and skip chapters.json generation."

The workflow in practice

Here’s what using this tool looks like for a real episode:

  1. After your session: Create the episode folder, jot down the tracklist and highlights in notes.json while it’s fresh.
  2. After editing the video: Update timestamps to match the final edit.
  3. Before upload: Ask your agent to run the builder on episodes/s2e04-warehouse/ (or run node episode-builder.js episodes/s2e04-warehouse/ in your terminal).
  4. During upload: Copy-paste from description.md, pick a title from titles.md, follow checklist.md.
  5. Post-publish: Use the social media snippets for cross-promotion.

Total metadata time: 2 minutes instead of 30.


Try it yourself

  1. Create an episode folder with a notes.json file (use real episode data or make one up).
  2. Talk your agent through the plan in plan mode and approve it to build the tool.
  3. Run the tool and review each output file.
  4. Copy the description into a test YouTube upload (or just a text editor) to verify formatting.
  5. Pick one customization and add it.

Key takeaways

  • Structured input, structured output — a JSON notes file produces consistently formatted metadata every time.
  • YouTube timestamps are critical — properly formatted timestamps auto-generate chapters, which boost watch time and viewer navigation.
  • Checklists prevent forgotten steps — the publishing checklist catches things you’d otherwise miss (end screens, premiere settings, cross-platform uploads).
  • The tool fits your workflow, not the other way around — you write notes after your session, update timestamps after editing, then generate everything before upload.
  • Templates make the next episode faster — once you have the structure, creating notes for a new episode takes 5 minutes.

KNOWLEDGE CHECK

Your YouTube description has a full tracklist with timestamps, but the first one is '0:45' because you trimmed the intro, and YouTube isn't creating chapters. What's likely wrong?


What’s next

In the next lesson, you’ll build The Moodmix Control Room: a private analytics dashboard that combines your Spotify for Artists and Shopify data to show where your listeners are, who’s buying merch, and where the opportunities are. Real business intelligence for an independent artist.

intellimetrics · Arlington, VA
© 2026 intellimetrics.net
Search lessons