Skip to main content

codex-continue

Expect-style wrapper that automatically sends a "Continue" message to Codex CLI after your usage limit resets or when the selected model is at capacity.

Quickstart:

pipx install codex-continue
codex-continue -- "/goal finish this task"

Background: you have a long task for codex, and a cheap Pro plan. Your task if for fun and you don't want to spend more money on it. But you also don't want to waste your unused plan quotas or your time waiting to click "Enter" when the limit refreshes. This script is for you.

This script waits for a message with the following format to show up:

You've hit your usage limit. Upgrade to Pro (https://chatgpt.com/explore/pro), visit https://chatgpt.com/codex/settings/usage to purchase more credits or try again at 11:42 AM.

When that happens, it parses out the time displayed, sleeps until it is reached and then wakes up and then automatically sends a message which defaults to:

Continue

which should be enough to send Codex on its merry way once again until the next quota is hit and the cycle continues.

Tested on Ubuntu 25.10, Python Python 3.13.7, codex-cli 0.133.0.

Installation

From PyPI:

pipx install codex-continue

Use latest master executable from a clone:

git clone https://github.com/cirosantilli/codex-continue
cd codex-continue
pipx install -e codex-continue

Usage

Run Codex through the wrapper:

codex-continue

Pass an initial prompt to Codex. /goal is recommended, Codex seems to work for much longer with it now:

codex-continue -- "/goal finish this task"

Customize the continuation text:

codex-continue --continue-prompt "Continue exactly where you left off"

When Codex prints a usage-limit message with either an absolute reset time or a relative duration, codex-continue keeps the same Codex process alive, waits until the reset time plus a grace period, then injects the continuation prompt and presses Enter.

It also handles:

Selected model is at capacity. Please try a different model.

For capacity errors, it waits five seconds and sends the continuation prompt in the same session, retrying each time the error appears. Customize this delay with:

codex-continue --capacity-retry-seconds 10

The usage-limit grace period does not apply to capacity retries. The selected model stays the same.

Publishing

Install the publishing tools once:

pipx install build
pipx install twine

Commit your changes, then run the release script from the repository root with a clean working tree:

./release

It runs the tests, increments the minor version and resets the patch version (for example, 0.1.1 → 0.2.0), updates all three version declarations, builds into dist/<version>/, and checks the artifacts. It commits the version bump as Release <version>, uploads only that version's wheel and source distribution, then creates an annotated Git tag named v<version>. It stops on failure and rejects existing tags. The commit and tag stay local; push them when ready:

git push
git push origin "v$(python3 -c 'from codex_continue import __version__; print(__version__)')"

Tests use the Python interpreter running release by default; set PYTHON=/path/to/python to use an environment containing the project's dependencies. The build uses the pyproject-build command installed by pipx.

To build and validate without uploading, committing, or tagging:

./release --no-upload

The bumped version remains in the source files if a build or check fails, and the release commit remains if the upload fails. To retry, or to publish after --no-upload, first commit any pending changes, then keep the current version:

./release --no-bump

PyPI does not allow replacing previously uploaded files, even after deletion. If an upload partially succeeded, upload only the missing artifact rather than rebuilding and replacing the published file.

Twine accepts the PyPI API token through TWINE_PASSWORD, with TWINE_USERNAME=__token__. For Bash, read the token without putting it in shell history:

read -rsp 'PyPI API token: ' TWINE_PASSWORD; echo
export TWINE_PASSWORD
export TWINE_USERNAME=__token__
./release

The same variables work for a direct upload of existing artifacts:

twine upload dist/codex_continue-0.1.1-py3-none-any.whl dist/codex_continue-0.1.1.tar.gz

The script defaults TWINE_USERNAME to __token__; without TWINE_PASSWORD, Twine can prompt for the token. Set TWINE_NON_INTERACTIVE=1 to fail instead of prompting when credentials are missing. See Twine's environment variable documentation.

Metadata

Release files for codex-continue 0.3.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for codex-continue 0.3.0
File Size Uploaded Provenance Actions
codex_continue-0.3.0.tar.gz 10.2 kB Without provenance Details for codex_continue-0.3.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for codex-continue 0.3.0
File Interpreter ABI Platform Actions
codex_continue-0.3.0-py3-none-any.whl Python 3 none any Details for codex_continue-0.3.0-py3-none-any.whl

Total release size: 18.0 kB

Release files / codex_continue-0.3.0.tar.gz

Download URL codex_continue-0.3.0.tar.gz
Size 10.2 kB
Tags Source
SHA-256 checksum
How to use checksums
00cb9f766dd1344d0d19c084d15eebbbaed767ffdf81953c80baeca955029d01
BLAKE2b-256 checksum
How to use checksums
ee008ce78b0bdd1c9b0222d63ed80865a4e0d7beb981197a96038cf1d3eabb03
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.14.4

Release files / codex_continue-0.3.0-py3-none-any.whl

Download URL codex_continue-0.3.0-py3-none-any.whl
Size 7.8 kB
Tags Python 3
SHA-256 checksum
How to use checksums
5361e394d28a8d244897f58c6d1cba72120401fa7d1cb4fbe6ec14bb4f8a9d82
BLAKE2b-256 checksum
How to use checksums
a09a2b818c1b8992f2c71d0a32162eba4931546e2cc85112e017f8b243d581ec
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via twine/7.0.0 CPython/3.14.4

Security for codex-continue 0.3.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Release history Release notifications | RSS feed

This release

0.3.0 This release

2 release files No provenance

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page