Audible Ingest

audible-ingest

Extract your Audible clips, bookmarks, and typed notes, transcribe the clip audio with OpenAI Whisper, and emit a tidy markdown file per book.

This is what the Audible app *should* let you do: get your own highlights out in a format you can search, sync, or paste into a note app.

What it does

For each book you ingest, you get a markdown file like:

# The Psychology of Money

**Author:** Morgan Housel
**Narrator:** Chris Hill
**ASIN:** B08D9TXF3H
**Runtime:** 5h 55m
**Clips:** 36
**Extracted on:** 2026-04-14

---

## Clips

### 1

**Timestamp:** 00:02:12 | **Duration:** 30s | **Created:** 2021-07-08

> A genius is the man who can do the average thing when everyone else around
> him is losing his mind.

**My note:** Reminds me of Buffett in 2008.

Book text comes from Whisper transcription. Your typed notes are preserved verbatim.

Install

git clone https://github.com/outdoorventures/audible-ingest.git
cd audible-ingest
bash scripts/setup.sh

`setup.sh` creates a Python venv at `~/.audible-ingest/venv/` and installs `audible`, `audible-cli`, and `httpx`. You also need:

• ffmpeg — brew install ffmpeg (macOS) or your distro's package manager. Used to decrypt and slice clips from the AAX file.
• OpenAI API key — export as OPENAI_API_KEY for the session. Needed only if you want Whisper transcriptions; skip by passing --metadata-only.

Claude Code skill (optional)

If you use [Claude Code](https://docs.claude.com/claude-code), symlink this repo into your skills directory and Claude will auto-trigger the pipeline on requests like "export my Audible clips" or "ingest *The Psychology of Money*":

mkdir -p ~/.claude/skills
ln -s "$(pwd)" ~/.claude/skills/audible-ingest

First-time auth

Audible's OAuth is browser-based and can't be scripted end-to-end, so auth is two steps:

~/.audible-ingest/venv/bin/python scripts/auth.py step1
# Open the printed URL, log in, copy the full r