Start here
Tokenome documentation
Tokenome links every line of AI-written code to the conversation that produced it. It runs on your own machine: it captures the conversations you already have with your AI tools, indexes them locally, and answers the question your commit history cannot, which is why.
What it captures
Tokenome reads conversations from the tools you already use and turns them into searchable segments stamped with their time, project, platform and model. Tool actions are captured too, not only the chat around them.
| Source | How it arrives |
|---|---|
| Claude Code | Read from ~/.claude/projects automatically, and at the end of each session if you install the plugin. |
| Claude Desktop and Cowork | Pulled from the claude.ai API with your own session key, when you set one. |
| ChatGPT | Export your data, unzip it, and drop conversations.json into the ChatGPT import folder. |
| Gemini | Drop a flat JSON export into the Gemini import folder. |
What stays local
On the free tier, all of it. Your conversations stay on disk under your home directory. Embeddings are computed on your own hardware by a model that ships inside the download, so no text is sent anywhere to be turned into vectors. The index is a search engine running on your machine. There is no account to create and no sign-in.
Nothing leaves your machine. No account. No cloud embeddings.
The one thing that does reach the network is the desktop app's update check against the public releases page, which carries no conversation data. The privacy page lists every file and every exception.
The same diagram in words
Your AI coding tools, Claude Code, Claude Desktop, ChatGPT and Gemini, write their transcripts to disk as you work. Tokenome reads them where they land.
On the same machine it splits the threads into turns, computes the embeddings on your own hardware, and indexes everything locally in Typesense. Code provenance links a line of code back to the conversation that produced it.
You reach the index through the app, the command line and a local web UI, and an agent reaches it over MCP. All of that is free, and none of it leaves the machine.
One gate crosses the boundary. When you opt a project in, its segments are redacted on the laptop and then mirrored to a team server your organization runs, where a teammate's search finds them with your name on them. Tokenome does not host that server.
Contents
- InstallThe macOS DMG, the Linux AppImage and deb, the command line, and the Claude Code plugin.
- First runWhat happens when you launch it, what gets indexed, where the data lives.
- Finding thingsSearch, filters, the web UI, and the command line that does the same job.
- Code provenancePoint at a line and read the conversation behind it, and what that cannot tell you.
- The journal and FAQA day of decisions with quoted receipts, and the running answer history.
- Agents and MCPThe tools an agent gets, connecting Claude Code, and what the plugin adds.
- Privacy and your dataWhat never leaves the machine, where the files are, how to delete or re-index.
- TeamStanding up the server, enrolling laptops, the share gate, team search.
- Admin consolePeople, devices, recovery, the audit log, retention, holds and seats.
- TroubleshootingThe handful of things that actually go wrong, and what to do about each.
The Team guides as PDFs
The two guides for standing up and running a team server are also PDFs, rendered from these pages so they cannot drift: Team server setup and Admin console. Each carries the page it came from and the date it was rendered.
Where to start
If you are installing it for yourself, read Install, then First run, then Code provenance. That is the whole product on one machine, and it is free.
If you are standing it up for a team, read Team and Admin console. You run the server; Tokenome does not host one for you. Team is in beta and free while the beta runs.