# How do you use a ChatGPT Dot in ChatGPT Space? Setup steps, shared to-do and handoff pages, and fixes when @dot doesn't answer or files won't open

A ChatGPT Space page works well as a shared notebook with your Dot: a running to-do list, a project tracker, a handoff page with Codex or a handover note. Six setup steps, what to do when an @dot comment gets no reply or the Dot can't open Space or Library files, why schedules belong in chat, and one message to start with.

What a Dot can do on a ChatGPT Space page, six setup steps, fixes for unanswered @dot comments, unreadable Space and Library files and schedules that won't attach, and one starter message.

People use Space pages with their Dot in four ways. Each one has a different weak spot, so it's worth knowing which one you're building.

### A running list the Dot keeps current

Dan Mallatt [keeps an Ideas & Projects page](https://x.com/dan_mallatt/status/2108240079391162806) that the Dot helps manage, in place of Notion or Todoist. Each day the Dot reviews the connected email, calendar, docs, and their conversations for commitments, deadlines, and project updates, flags stale items, tracks what's still waiting on someone, and adds any idea dropped into the chat without switching apps.

Jam (@jamout.ai), who posts about AI and marketing, [describes a similar setup on TikTok](https://www.tiktok.com/@jamout.ai/video/7694014345088421134): a Dot named Candy organizes the pages in Jam's Space, and a project tracker page updates every morning at 6 a.m.

goryugo [uses a Space page the same way](https://knowledgestuck.substack.com/p/obv026): the Dot writes the to-do list and the steps for each job, both of them edit the same page, the Dot carries on with what it can do alone, and the items that need a decision are settled in conversation.

The weak spot is trust in the list itself. Lee Shand's To-do Page refreshed at 6 a.m., and one item read as more settled than the conversation behind it; that case is on [what goes wrong once a daily brief is scheduled](/use-cases/morning-brief/). Similar pages show up for household lists in [family and health](/use-cases/family-and-health/) and for itineraries built from booking emails in the [travel planner](/use-cases/travel-planner/) guide.

### A handoff page between your Dot and other agents

Keith (@icautopart) [posted a setup marked as tested](https://x.com/icautopart/status/2107296660540084709) that links a Dot, a desktop Codex task, and a Space. The steps: open the Codex task in the project folder, create a coordination page for goals, progress, permissions, and next steps, tell the Dot which task and Space to use, and explicitly authorize the two tasks to message each other. Then check the link: the Dot sends a unique test phrase that the Codex task has to answer, and the Codex task reads the page, adds a short acknowledgement, and confirms it saved. The post notes the setup does not grant permission to restart services or make other operational changes.

ヤッテミストひで (Hide), who learns AI tools on his own and isn't an engineer, [used a Space page as the meeting point between Codex on his home PC and his Dot](https://note.com/hide_taxnote/n/nff04fcec8460) (note, 2 October 2026, in Japanese). Copying text between the two chats had become a chore, so Codex wrote its research results to a Space page, and the Dot read the report, summarized it and suggested what to look into next. Hide talked the suggestions over with the Dot, partly from his phone while he was out, and the Dot added the agreed follow-up request to the same Space. He then messaged Codex from his phone, "I've written an additional request in the Space, please check it and carry on," and Codex picked up the extra research. The step he hadn't managed yet was having the Dot start the Codex work itself, so he still sent that last message by hand.

Andric (@astralwave) [has a Dot post regular updates to a Space page](https://x.com/astralwave/status/2107717615229825288) while it coordinates with another agent over Slack, with a separate friction log where both agents record what got in their way. The verdict a few days in: better than before, but they "drop the ball a lot." Alex Finn's project-manager skill, which keeps its evidence in a project Space, is covered in [Dots with Codex](/use-cases/codex-orchestration/).

### A handover page before you rebuild a Dot

Fotios, on the OpenAI Community forum, [hit the Slack rename problem](https://community.openai.com/t/dots-broken-when-trying-to-rename-on-slack-internal-server-error/1404100/6) (10 October 2026): after removing the Dot from Slack to rename it, it couldn't be added back. The fix was to delete the Dot and create it again from the web, and before deleting it, Fotios had it write a handover document in a Space first. A Space page outlives the Dot that wrote it, so it's the obvious place to keep what a new Dot would need: its jobs, schedules, connected apps, open threads and the rules you gave it. The full Slack story is on the [Slack integration page](/integrations/slack/).

### Notes for the agent that people can't find

brandon galang [likes Space as persistent storage for agents](https://x.com/brandon_galang/status/2107939281193128310) but can't find what the Dot creates: the All view shows only parent pages, so nested pages are hard to reach, and Space has become a note surface for the agents only. Roman (@khudonogov) [asked a Dot to find every page from a test run](https://x.com/khudonogov/status/2107134820690063520); it found them all with context, but took about three minutes. Asked next to move four personal Pages into a Space created earlier, it said the move wasn't supported, offered copies instead, and moved nothing. Antony (@idsutong) [reports](https://x.com/idsutong/status/2107789610529214812) that the Dot and ChatGPT can open files in the Space but a Codex session can't, so the files get copied to Google Drive.

Source: https://gptdot.net/chatgpt-space/
