Chronicle KeeperDocs
Docs / Getting started

Getting started

A local-first worldbuilding workspace for your campaign — an AI-assisted wiki, maps and a timeline, with session recordings folding in as notes. Here's how to get going.

Chronicle Keeper is a desktop app for building and running your tabletop world. Your campaign is a wiki of Markdown pages you own — NPCs, places, factions, lore — linked together, pinned on maps and plotted on a timeline, with the Keeper (an AI that has read all of it) to answer questions and keep the canon current. It also turns Craig Bot session recordings into clean notes that fold back into your world. Everything heavy runs on your own machine; you only reach the internet to download the app, the speech model (once), and to call a cloud LLM if you pick one over a local model.

New to all this?

You don't need to be technical. If you can install an app and copy-paste a key, you can run Chronicle Keeper. Start with Worlds & pages to learn the wiki, or the recordings guide for session notes.

A Chronicle Keeper world overview
A world in Chronicle Keeper — the codex, the party and the story so far in one place.

What you'll need

  • A Mac, Windows PC, or Linux machine.
  • An LLM for the Keeper and the writing — either Ollama running locally (free) or an API key from a cloud provider.
  • Optional, for session notes: a Craig Bot recording — the ZIP it emails you, one audio track per speaker. (How Craig works →)

Quick start

  1. Install the app

    Grab the installer for your operating system from the Releases page. First launch shows a one-time "unknown developer" prompt — that's expected.

    Download   Per-OS install steps →

  2. Choose your LLM in Settings

    Open Settings and either point it at a local Ollama server or paste a cloud API key. Keys never leave your machine.

    Full LLM setup guide →

  3. Create your world

    Start a fresh world (a portable folder on your disk) or open an existing one — even an Obsidian vault. Add your first pages for the people and places that matter, and ask the Keeper anything.

  4. Optional: add a session recording

    Start a session and drop in your Craig ZIP, then label which track is which player. On the very first transcription the speech model (Parakeet TDT v3) downloads once; after that every track is transcribed on-device.

  5. Summarize & export

    Generate the summary with your chosen LLM, give it a read, then export to Markdown with Obsidian frontmatter.

    See the whole workflow →

First-run downloads

The app itself is small, but the speech model is a few hundred MB and downloads once on your first transcription. After that it's reused and you can work fully offline.

Where to go next