What it does
Parses an episode transcript to identify topic shifts and builds a timestamped outline, pulls out quotable lines and any resources or links mentioned in conversation, then drafts a publish-ready show notes post in the site's existing format. Saves the manual work of relistening to mark timestamps by hand.
Install
This skill isn't packaged in an installable registry. It's a plain SKILL.md file you can copy directly. Copy it from the SKILL.md tab and save it at the path for your assistant:
Use it in your AI assistant
Claude Code
.claude/skills/podcast-shownotes-writer/ (or ~/.claude/skills/podcast-shownotes-writer/ for all projects)
GitHub Copilot (VS Code)
.github/skills/podcast-shownotes-writer/
OpenAI Codex
.codex/skills/podcast-shownotes-writer/
Cursor
.cursor/skills/podcast-shownotes-writer/ (or ~/.cursor/skills/podcast-shownotes-writer/ globally)
Compatibility
Any WordPress site publishing podcast episodes. Requires a transcript, or audio the agent can transcribe first.
The complete SKILL.md for this skill:
---
name: podcast-shownotes-writer
description: Use when a user has a podcast episode transcript and wants show notes, timestamps, or an episode summary written for publishing.
---
# Podcast Show Notes Writer
## When to use
A podcast episode has been recorded and transcribed (or the audio can be transcribed), and the user needs a show notes post ready to publish alongside it.
## Procedure
1. Read the full transcript, or transcribe the audio if no transcript exists yet.
2. Identify natural topic shifts in the conversation and mark each with an approximate timestamp.
3. Pull out 2-4 quotable lines that capture the episode's most interesting points, attributed to the correct speaker.
4. Note any resources, tools, books, or links mentioned by name during the conversation, listed separately for the notes.
5. Draft the show notes post: a short episode summary, the timestamped topic outline, pulled quotes, and the resource list, matching the site's existing show notes format.
6. Flag any timestamp where the transcript is ambiguous or garbled so the user can spot-check against the audio before publishing.
## Guardrails
- Don't attribute a quote to the wrong speaker when the transcript doesn't clearly label who's talking, flag it as uncertain instead.
- Keep timestamps approximate unless verified against the actual audio, mislabeling them is worse than leaving them out.