Skip to content

Releases: dmxj/tokenbook

Release list

TokenBook v0.1.0

Choose a tag to compare

@dmxj dmxj released this 21 Sep 07:22

First release. A local macOS app that indexes the conversation history of every AI coding agent
on your machine into one searchable database.

What's in it

  • Six agents, four on-disk formats — Claude Code, Codex, OpenCode, and the Gemini CLI family
    (Gemini CLI, Qwen Code, iFlow), with auto-detection of which layout a folder uses.
  • One search box over everything — ⌘K runs an FTS5 query across every agent and project.
    Scope to the current session, one agent, or all of them; filter by time range. A switch brings
    thinking blocks, tool calls and intermediate output into scope.
  • Browse by project directory — sessions grouped by the working directory they recorded, tabbed
    per agent, with a subdirectory roll-up.
  • Pin and favourite — pin projects and sessions; pins float above whatever sort you pick.
  • Export — a whole session or a single turn, as Markdown or JSON, at full fidelity.
  • Live sessions — a session an agent is still writing to re-indexes within a second.
  • English, 简体中文, 日本語, 한국어 UI, four themes. Conversation content is never translated.

Install

Build from source (recommended — it's a two-command build):

git clone https://github.com/dmxj/tokenbook.git && cd tokenbook
pnpm install && npm run install:mac

Or grab TokenBook-0.1.0-arm64.dmg below.

Caveats

  • The build is not code-signed — there's no paid Apple developer certificate behind this. On
    first launch macOS will say the developer cannot be verified: right-click the app → Open, or allow
    it under System Settings → Privacy & Security.
  • macOS only, and only tested on arm64. It's an Electron app so other platforms are probably
    reachable, but they are untested.
  • It reads your agent history read-only and makes no network requests. Deleting the index only costs
    you a re-index.
842 sessions, 6,466 turns → 583 MB index, searches in single-digit milliseconds