Switch from Codex to Task Machine

Codex keeps coding work in app, terminal, editor and cloud chats. Task Machine gives people and agents shared Tasks with an Inbox for plans, questions and review.

Prefer the side-by-side comparison?

Why people switch

  • Local schedules need your computer

    Codex project schedules need the computer powered on and the desktop app running. That matters when a nightly repository check depends on local files.

    learn.chatgpt.com/docs/automations.md
  • Usage varies by job

    Codex says similar-looking tasks can consume different amounts of your allowance because models, context and tool use affect usage. Published message estimates are not fixed limits.

    learn.chatgpt.com/docs/pricing.md
  • Guidance needs a durable home

    Codex tells you to keep required team guidance in AGENTS.md or checked-in documentation. Its local memories are a recall layer, not the only place for rules that must always apply.

    learn.chatgpt.com/…/memories.md

What maps to what

Every answer links to its source and shows when we last checked it. The oldest check on this page was November 15, 2026.

What each part of Codex maps to in Task Machine, with the source and check date of each answer
Area In Codex In Task Machine
Tasks In Codex

Each Codex cloud task has its own workspace. You inspect changed files and checks, request follow-up changes, then commit or open a pull request.

learn.chatgpt.com/docs/cloud.md
In Task Machine

Every job is a Task that a person or an agent owns. The Task keeps its plan, discussion, decisions, runs and history together, so people and agents can hand work to each other on the same record.

taskmachine.io/docs/tasks
Workflows In Codex

Skills package reusable instructions and resources for jobs such as documentation review. Plugins can bundle skills, connected tools and runtime hooks.

In Task Machine

A Workflow repeats explicit steps with branches, human questions and approvals. Verifier checks judge a result against its requirements and send failed work back with feedback.

taskmachine.io/docs/workflow-builder
Approval In Codex

Sandbox settings control what Codex can access. Approval policies control when it asks before acting, such as leaving the sandbox or running a command that needs network access.

In Task Machine

Decisions arrive in your Inbox with their context. Set each agent's actions to Ask first or Automatic, and put an explicit Human approval step before publishing a report.

Code review In Codex

Codex reviews GitHub pull requests on request or automatically. It follows the applicable code review rules in root and nested AGENTS.md files.

learn.chatgpt.com/…/github.md
In Task Machine

Internal code review checks completed changes against enabled repository rules. Rule discovery proposes rules after screening them against violating and valid examples, and you decide which to accept.

taskmachine.io/docs/repositories
Execution In Codex

Codex CLI uses your local repository and installed tools. Codex Cloud can keep working while your computer sleeps.

In Task Machine

Run agents in Managed Cloud with nothing to install, or connect a Local Worker for files, tools and accounts on a computer you control. You choose where each agent runs.

Pricing In Codex

USD documentation lists Plus at $20 a month and Pro from $100 a month. API-key usage is billed separately at API rates and excludes cloud-based features such as GitHub review.

In Task Machine

Usage based, from $29 a month before tax. The whole payment becomes usage that your workspace shares, with no base fee.

taskmachine.io/pricing

A fact we cannot find or cannot check is left out, and prices can change after the date shown. Task Machine publishes this comparison and is an interested party. Codex is named only to identify it, and there is no relationship between the two companies. Has something changed? Tell us.

You may stay with Codex if

You may stay with Codex if you want an agent beside your source while you work. Its IDE extension brings open files and selected code into the prompt and lets you review edits in place. The extension supports Visual Studio Code and compatible editors such as Cursor and Windsurf. Moving shared work into Task Machine does not require abandoning that editing loop.

What you give up

Task Machine's docs do not describe an equivalent editor extension or a Codex conversation importer. You rebuild instructions, reconnect tools and test review boundaries rather than carrying over a running chat. For repository work, put run instructions in TAMA.md and review the enabled repository rules separately.

There is no trial. Your first payment funds usage, and unsuccessful attempts can still consume it. Compare one representative job before moving a large queue, using Usage and fees.

How the switch works

  1. Keep the code and context first. In Codex CLI, wait for the active turn to finish, then use /export to save the conversation as Markdown. Review and commit or open a pull request for the cloud changes you want to preserve. Copy the prompts, schedules and skill resources for the job you are moving.
  2. Create your workspace, add the repository, and configure Connectors for the services the job needs. Keep Git access separate from the Connector used for pull-request actions. Upload selected reference material to the Library, after removing secrets and unrelated conversation.
  3. Create one Task with the intended result, repository, owner and acceptance requirements. Rebuild a repeated skill-driven process in the Workflow builder, with a verifier where an output check would help.
  4. Start at Supervised autonomy. Review the plan and a complete result before allowing external actions. Use Managed Cloud or a Local Worker according to where the required files and tools live.
  5. Pause the original Codex schedule for the selected job before enabling its replacement. Set Task Machine's schedule and timezone, then inspect the first occurrence before increasing volume.

Common questions

Can I keep Codex for interactive coding?
Yes. Keep its terminal or editor workflow and move one shared responsibility, such as a weekly repository report, into Task Machine first.

What happens to my chats and memories?
Codex CLI documents Markdown conversation exports. Its local memory files can be inspected, and its docs warn you to review them for secrets before sharing. Keep those archives separately and move selected guidance into the Library. Task Machine's docs do not describe importing Codex chats or its memory store as live state.

How do the costs compare?
Codex subscriptions and API billing cover different usage from Task Machine. Task Machine is usage based, from $29 a month before tax, with the whole payment shared as workspace usage, no base fee and no trial. Test the same job rather than assuming equal task counts. You keep 100% of your revenue, Task Machine takes no cut and never custodies your accounts. See pricing.

Does Task Machine run Codex as its worker?
Task Machine agents use tamacode, its own coding agent. A Local Worker can use a ChatGPT subscription you sign in to, but that is model access rather than hosting the Codex app, CLI or extension.