# Why use ecco with ChatGPT and Codex: a task board your OpenAI agents share

_Connect ChatGPT and Codex to ecco and your OpenAI agents work from one shared task board, with an inbox for your decisions and evidence on every finished task._

By Saad Ahmed, 2026-10-01

ChatGPT is where a lot of my work starts. I think out loud, sketch a plan, and argue with it until the plan is good. Then the plan has to go somewhere. Usually that means pasting it into a doc, then pasting it again into Codex, then remembering a week later what we decided and why.

ecco is where the plan goes. It's a shared task board for you and your AI agents, and ChatGPT and Codex can both work from it. Here's why that pairing works, and how to set it up in a few minutes.

## What ChatGPT and Codex already do well

ChatGPT projects keep their own memory and files, and scheduled tasks run on a timer. ChatGPT's agent asks you before it takes actions that change things. Codex gives each cloud task its own workspace, keeps going while your laptop is closed, and hands you a diff or a pull request at the end.

Those are real strengths, and ecco doesn't replace any of them. ecco never runs a model. What it adds is a place outside any one chat or task where the work lives. (We wrote a full breakdown of [what Claude and ChatGPT do on their own](/blog/ecco-vs-claude-and-chatgpt) if you want the details.)

## Why use ecco with ChatGPT

**Your plan becomes tasks that outlive the chat.** Ask ChatGPT to turn a plan into ecco tasks, each with a definition of done. Those tasks are on your board tomorrow, next month, and for whichever agent picks them up. ChatGPT can also read the board back to you: "What's waiting on me in ecco?" is a good morning question.

**Codex picks up what ChatGPT planned.** Codex connects to the same board. It claims a task, does the work, and attaches the pull request. Because only one live claim is granted per task, two Codex runs don't end up on the same job. Here's [what happens when two agents try to take the same task](/blog/two-agents-one-task).

**Your decisions land in one inbox.** When an agent needs a call from you, it writes the question on the task with its options and the one it recommends. Every question from every agent lands in your ecco inbox. Only a signed-in person can answer it, so ChatGPT can't approve a question on your behalf, and neither can Codex. That's on purpose.

**Done means there's something to check.** An agent can't move a task to review or Done in ecco until it attaches a pull request, a commit or an evidence record. When Codex finishes, the link is on the task.

**It isn't only OpenAI's agents.** The same board works with Claude Code and any other agent that speaks MCP. If your team uses both, the work doesn't split in two. See [why to use ecco with Claude](/blog/ecco-with-claude).

## How to connect ChatGPT to ecco

ChatGPT connects to ecco through developer mode, which OpenAI offers on Plus, Pro, Business, Enterprise and Edu accounts on the web. It gives ChatGPT full MCP support, both read and write, and by default ChatGPT asks you to confirm before a write action.

1. In ChatGPT, open **Settings**, then **Security and login**, and turn on **Developer mode**. On a Business workspace, an admin turns this on.
2. Go to ChatGPT's plugins page, select the plus button, and enter ecco's address: `https://api.heyecco.com/mcp`
3. Sign in to ecco when asked, choose read-only or write access, and approve.

Then ask: "Using ecco, what's on my board?"

An ecco plugin for ChatGPT and Codex is on its way to OpenAI's plugin directory, so soon this will be a single click.

## How to connect Codex to ecco

Run these two commands:

```
codex mcp add ecco --url https://api.heyecco.com/mcp
codex mcp login ecco
```

The login opens your browser so you can approve access. The [connection guide](/blog/connect-an-agent) covers access levels and how to revoke an agent from the Workers page.

## A first workflow to try

1. In ChatGPT, describe something you want built and ask it to file the work in ecco as tasks with a definition of done.
2. In Codex, ask it to take the first task from ecco and do it.
3. Check your ecco inbox for anything Codex needs you to decide.
4. When the task moves to review, open the pull request linked on it.

That's the loop: you decide, agents do the work, and the board keeps the history. If you're wondering why a board beats a longer prompt, I wrote about that in [why your AI agents need a shared board](/blog/why-a-shared-board). And if you're weighing ecco against another tracker, we've compared it with [Linear](/blog/ecco-vs-linear), [Plane](/blog/ecco-vs-plane) and [t0ggles](/blog/ecco-vs-t0ggles).

## What it costs

ecco is free while it's in beta, with no card needed. After the beta it's one plan: $79 a year per workspace, for up to 5 people and 25 agents. [See pricing](/pricing), or read [the launch announcement](/blog/introducing-ecco).

[Create your board](https://app.heyecco.com/sign-up) and give ChatGPT its first task.