Your AI tracks its own issues,
and closes them as it works
One issue space your sessions and agents share.
Free for everyone during the introduction. Connect your tool, sign in, and start.
- Work that outlasts a conversation is kept, not lost when a session ends
- Sessions working at once see each other's claims and leave them alone
- What only you can decide waits for you, and stays open until someone acts
- For code, research, support, scheduled runs, writing, and agent teams
Connect Claude or ChatGPT
Claude
ChatGPT
Claude Code
Codex
tzTrack #
Turn Zero Track (tzTrack) is an issue space your AI keeps for itself across sessions. Your AI raises an issue for anything that must outlast its conversation, and closing the issue is the check-off. Other sessions and agents share the space, several at once, so work one session leaves is there for the next. tzTrack is free for everyone during the introductory period.
Start here #
In Claude, the desktop app or claude.ai, open Settings, then Connectors, and add a custom connector with this address, where your plan or organization allows custom connectors. In ChatGPT, add it as a custom connector in its settings, with developer mode on, where your plan offers it:
https://tztrack.ai/mcp
Then sign in with your Turn Zero account in your browser.
Using Claude Code or Codex? Each adds the server with one command in a terminal. In Claude Code:
claude mcp add --transport http tztrack https://tztrack.ai/mcp
Next, type /mcp in Claude Code, choose tztrack, and finish the sign-in with your Turn Zero account in the browser. In Codex, run this command:
codex mcp add tztrack --url https://tztrack.ai/mcp
Codex then opens the browser for the sign-in. Where the browser does not open, run codex mcp login tztrack.
Connect your AI tool gives every step, the check, and the way to connect any other AI tool. A script or a CI job with no AI tool follows the Quick start instead.
What it is for #
For you, tzTrack keeps what your AI must not forget. Work it does not finish now, work another session or agent picks up, and anything you must see or do stay in the space. They stay there when a session ends or is archived. You ask your AI what is open, and it answers from the space.
For your AI, the space is where anything that must outlast this conversation goes. Your own task list holds the steps of this conversation. tzTrack holds the rest, one point per issue, shared with every other session working the same space.
The guides show it at work. Coding is one use among several:
- Track issues while you code
- Keep research threads open
- Follow up customer conversations
- Record what a scheduled run finds
- Track a writing project
- Coordinate an agent team
- Share a space
What it costs #
tzTrack is free for everyone during the introductory period.
How the pages fit together #
- Quick start gives the one line for each AI tool, and the REST API's steps for a script with no AI tool.
- Connect your AI tool adds the server and signs in.
- Name your space in a setting file makes every session of a project use the same space.
- Issues and kinds explains what an issue holds, its states, its kinds, and its outcomes.
- Waiting and claims explains what an issue waits for and how a claim keeps sessions apart.
- Coordinate an agent team shows how sessions working at once claim issues and keep out of each other's way.
- Share a space adds people to your space by their Turn Zero accounts.
- Keep a local cache keeps a copy of the space's issues on your computer, which your AI searches with its own file tools, in a tool that runs commands.
- The tools lists every tool the server serves.
- REST API works the space from a script or a CI job over HTTPS, with the same token.
- The setting file gives the file's one member.
The served version #
This manual describes tzTrack 0.1.0, the server at https://tztrack.ai/mcp.
What it does not do #
- An issue has no sub-items and no checklist. An issue is the item, and closing it is the check-off.
- There is no command to install and no plugin. Your AI tool connects to the server at its address, and that is the whole setup.
- tzTrack has no screens of its own. Your AI reads and writes the space for you.
- In Claude or ChatGPT, tzTrack keeps no setting file and no local cache. Both are files on your computer, which only a tool that runs commands, such as Claude Code or Codex, can use. Your AI reads the space live instead.
- An issue stays open until a session closes it with an outcome. The one exception is an issue left waiting for evidence with no new report for 90 days, which closes as
unexplained. - The server's instructions to your AI are guidance, not a contract. The contract is the one The tools links.
Related #
The glossary defines the words this manual uses. What happens without asking says what your AI does in the space on its own and what it asks you first. Refusals lists every refusal the server answers, with its cause and its remedy.
Machine surfaces of this manual #
This manual is published at https://turnzero.ai/tztrack/ with the four machine files its build writes:
- The index is at
https://turnzero.ai/tztrack/llms.txt, with one line per page giving its title, address, and description. - The whole text is at
https://turnzero.ai/tztrack/llms-full.txt. - The manifest is at
https://turnzero.ai/tztrack/docs.json. - The Markdown copy of each page is at the page's address followed by
index.md.
Only a Turn Zero account that holds tzTrack can read these pages and files. The readers:
- A browser signed in to such an account can read the pages and the files.
- A request with a bearer token of such an account gets the machine files, but no page.
- A tool connected to Turn Zero Cloud under such an account can read the index as the resource
docs://tztrack/llms.txtand the whole text asdocs://tztrack/llms-full.txt. It can also read them through theread_documentationaction. - Every other request gets a not-found response, the same response as for a path the site does not have.