assafkip

Notebooklm Daily Podcast — AI skill for Claude Code

AI community

A daily AI-news podcast that runs itself.

How to install Notebooklm Daily Podcast

This entry records only its repository, not the path inside it, so there is no exact command to give. Open assafkip/notebooklm-daily-podcast and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Notebooklm Daily Podcast does

A daily AI-news podcast that runs itself. Harvest your sources, Claude curates, NotebookLM makes the audio. Clone it, point it at your own feeds, it's your show. Config-driven, no brand lock-in.

Alternatives in AI

  • Geo SEO Claude — GEO-first SEO skill for Claude Code 3.6k ★
  • Generative Media Skills — Multi-modal Generative Media Skills for AI Agents (Claude Code, Cursor, Gemini CLI) 3k ★
  • Qwen Audio Agent — A realtime voice runtime that keeps Agents talking, working, and present 2.3k ★

README

notebooklm-daily-podcast

A daily AI-news podcast that runs itself. It harvests news from sources you choose, has Claude pick the few stories worth airing, makes sure nothing repeats, turns them into a two-host audio episode with Google NotebookLM, and (optionally) emails subscribers and publishes a podcast RSS feed.

Clone it, point it at your own sources, and it is your show. Nothing here is tied to anyone else's brand. Every choice that makes it *yours* lives in one config file.

What it does (the daily chain)

auth gate -> harvest sources -> Claude curates the best 4-5 -> dedup (no repeats)
          -> build a brief -> NotebookLM makes the audio -> [optional] deliver
          -> record what aired so tomorrow stays fresh

The default setup stops at the audio file. Delivery (email + podcast feed) is opt-in, so your first run needs zero hosting and zero subscriber list.

Requirements

  • macOS or Linux with python3 (stdlib only, no pip packages for the core), bash, and curl.
  • Claude Code (the claude CLI) with an Anthropic API key. This is the curation brain. https://docs.claude.com/claude-code
  • NotebookLM CLI (nlm, from notebooklm-mcp-cli) logged into a Google account. This is the audio engine.
  • (optional, for delivery) a Supabase project + a Resend account.
  • (optional, for extra sources) an Apify token for X/Twitter. Reddit needs no token: it reads the free Arctic Shift archive mirror.

Setup (about 10 minutes)

# 1. configuration
cp config.example.json config.json     # your show name, audience, host persona
cp sources.example.json sources.json   # where the news comes from
cp .env.example .env                   # your API keys

# 2. install the two engines
#    Claude Code:  https://docs.claude.com/claude-code
#    NotebookLM:   the `nlm` CLI, then `nlm login` (sign into your Google account)

# 3. prove it before you trust it (offline, no creds)
bash tests/run_selftests.sh

# 4. run it
bash run_da