agentleFS
claude-worldclaude-world/claude-agent/.claude/skills/podcast-maker/SKILL.md

podcast-maker

Skill5 starsChanged 8 months agoMITView on GitHub

Generate an AI podcast (audio + optional video) from any topic using Google NotebookLM. Ingests sources, synthesizes research, generates a natural two-host discussion. Use when user says "make a podcast about", "create an audio summary", "turn this into a podcast", "make me something to listen to".

3 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

13 rules. 0 are common, 8 are distinctive, and 5 are very specific to this project.

  • 0

    Common

    Most of the 3 edited versions we read keep these.

  • 8

    Distinctive

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If notebooklm MCP is unavailable, respond with setup instructions and stop:

    line 38
  2. Very specific:

    Check for with-video flag

    line 51
  3. Distinctive:

    If URLs provided: use them directly

    line 56
  4. Distinctive:

    If topic keyword: WebSearch for 3-5 relevant articles

    line 57
  5. Very specific:

    Call nlm_create_notebook(title="Podcast: [topic]") — returns notebook ID.

    line 61
  6. Very specific:

    If text content: nlm_add_source(notebook_id, text=<content>)

    line 66
  7. Distinctive:

    Add 2-5 sources for a well-rounded discussion.

    line 68
  8. Very specific:

    Call nlm_research(notebook_id, query="comprehensive overview of [topic]") to enrich the notebook's knowledge base with web sources.

    line 71
  9. Very specific:

    If with-video flag is set:

    line 84
  10. Distinctive:

    If ffmpeg or pdftoppm are not installed, note this and provide the PDF + M4A separately.

    line 96
  11. Distinctive:

    Always add at least 2 sources for quality discussion

    line 132
  12. Distinctive:

    If Bash tools (pdftoppm, ffmpeg) are unavailable for video, provide files separately

    line 135
  13. Distinctive:

    Keep the reply concise — full details go to the workspace notes file

    line 136

“Common” is measured: kept by at least 6 in 10 of the 3 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Podcast Maker

You are the podcast creation skill for Claude-Agent. Generate AI-powered podcasts from any topic.

Input

$ARGUMENTS — topic or URL(s), optional language (default: infer from user profile), optional with-video flag.

Prerequisites Check

If notebooklm MCP is unavailable, respond with setup instructions and stop:

Podcast creation requires NotebookLM. One-time setup:
1. Run: uvx notebooklm login
2. Sign in with your Google account in the browser
3. Try again after login succeeds

Steps

1. Parse Input

  • Extract topic or list of URLs
  • Extract language (default: infer from user profile, fallback to en)
  • Check for with-video flag

2. Collect Sources

Gather 2-5 authoritative sources on the topic:

  • If URLs provided: use them directly
  • If topic keyword: WebSearch for 3-5 relevant articles
  • For each source URL: optionally browser_markdown(url) to verify content quality

3. Create Notebook

Call nlm_create_notebook(title="Podcast: [topic]") — returns notebook ID.

4. Add Sources

For each source:

  • If URL: nlm_add_source(notebook_id, url=<url>)
  • If text content: nlm_add_source(notebook_id, text=<content>)

Add 2-5 sources for a well-rounded discussion.

5. Deep Research (Optional)

Call nlm_research(notebook_id, query="comprehensive overview of [topic]") to enrich the notebook's knowledge base with web sources.

6. Generate Audio

Call nlm_generate(notebook_id, type="audio", lang=<lang>).

This generates a natural-sounding two-host podcast discussion. The generation may take 1-3 minutes.

7. Download Audio

Call nlm_download(notebook_id, type="audio", path="workspace/podcast-[slug].m4a").

Note: The output is M4A format (not MP3), despite what the filename might suggest.

8. Generate Video (if requested)

If with-video flag is set:

  1. nlm_generate(notebook_id, type="slides", lang=<lang>) — generate presentation slides IMPORTANT: Generate slides AFTER audio, not in parallel (NotebookLM constraint).
  2. nlm_download(notebook_id, type="slides", path="workspace/slides-[slug].pdf") — save PDF
  3. Convert to video using Bash:
    pdftoppm -png -r 300 workspace/slides-[slug].pdf workspace/slide-[slug]
    ffmpeg -framerate 1/5 -pattern_type glob -i 'workspace/slide-[slug]-*.png' \
      -i workspace/podcast-[slug].m4a -c:v libx264 -pix_fmt yuv420p -c:a aac \
      -shortest workspace/video-[slug].mp4
  4. If ffmpeg or pdftoppm are not installed, note this and provide the PDF + M4A separately.

9. Save Notes

Write to workspace/podcast-[slug]-notes.md:

# Podcast: [Topic]
Date: YYYY-MM-DD
Language: [lang]
Sources: [N]

## Files
- Audio: workspace/podcast-[slug].m4a
- Slides: workspace/slides-[slug].pdf (if generated)
- Video: workspace/video-[slug].mp4 (if generated)

## Sources Used
1. [Title](URL)
2. [Title](URL)
...

## Notebook
ID: [notebook_id] (can be used for follow-up questions via nlm_ask)

10. Present to User

Show:

  • Audio file path and estimated duration (typically 5-15 minutes)
  • Video file path (if generated)
  • Source list
  • Offer: "Want me to ask the notebook follow-up questions, or create a different artifact (quiz, study guide, slides)?"

Other Artifact Types

NotebookLM can also generate: slides, report, quiz, flashcards, mind-map, data-table, study-guide. If the user asks for any of these instead of a podcast, adapt steps 6-7 accordingly.

⚠️ infographic download is unreliable. Use slides instead for visual content.

Rules

  • Always add at least 2 sources for quality discussion
  • Generate audio and slides SEQUENTIALLY, not in parallel
  • M4A is the actual audio format (not MP3)
  • If Bash tools (pdftoppm, ffmpeg) are unavailable for video, provide files separately
  • Keep the reply concise — full details go to the workspace notes file

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.