Turn a subject into a publication worth keeping. A considered workflow for words, photographs and the space between them.
Bring it into Codex ↗A place. A person. A point of view. Give the publication a reason to exist before giving it a look.
Then build the issue around it: a clear promise, a thoughtful sequence and images that do more than fill a page.
Editorial Studio gives Codex a repeatable editorial process. It asks what matters, develops two creative directions and keeps you involved at the decisions that shape the publication.
Define the subject, purpose and audience. Set the tone, share a reference or choose no example. The brief becomes the standard for every decision that follows.
OUTPUT / A CLEAR BRIEFUse the skill for editorial judgment.
Add the local application for production.
Instructions and references that guide Codex through planning, editing, brand development, photography storytelling and whole-issue review.
A separate Python application for organising features, reviewing images and exporting PDFs. Runs on your own device.
A4 portrait. Up to 12 magazine feature pages or 3 onepager features. RGB PDF output. High-end custom proof spreads are a separate, explicitly approved phase. No claim of CMYK, bleed or press-ready certification.
The source repository is private.
You need GitHub read access to install.
Use an account with read access. The local installer needs usable Git credentials or a supported GitHub token. Browser sign-in alone may not authenticate local Git.
$skill-installer install https://github.com/hndrk-020/editorial-studio/tree/main/.agents/skills/editorial-studioThe bundled installer normally uses ~/.codex/skills. It refuses to overwrite an existing skill folder; back up local edits before a deliberate update.
Start a new turn; restart Codex if needed. Use the verification prompt below and check the reported skill path.
Sign into GitHub with repository access. Open the repository, choose Code → Download ZIP on main, then Extract All. Do not run scripts inside the ZIP preview.
Double-click install_codex_skill_windows.cmd. It copies the full skill to %USERPROFILE%\.agents\skills\editorial-studio. No administrator access or Python is needed for this copy.
Existing installation? Back it up before approving replacement. The script overlays files; it does not remove obsolete files.
Test-Path "$env:USERPROFILE\.agents\skills\editorial-studio\SKILL.md"Run this in PowerShell. Expect True, then start a new Codex turn and use the prompt below to verify loading.
$editorial-studio Confirm you can load the skill and its editorial planning and brand system references. Tell me the loaded SKILL.md path. Do not start a publication yet.Keep the complete repository and install Python 3.11 or newer. The first launch requires internet to install dependencies.
Double-click run_windows.cmd in the extracted repository.
chmod +x run_mac_linux.sh && ./run_mac_linux.shOpen http://127.0.0.1:8765. Keep the terminal running; Ctrl+C stops the application. Keep it on localhost. Your publication files and exports remain local unless you choose to share them.
Keep one intended user-wide copy. The Windows installer uses ~/.agents/skills; the bundled Skill Installer normally uses ~/.codex/skills. A project-local copy can also exist. Ask Codex for the loaded path before changing anything.
Back up the whole existing skill folder outside active skill directories. Update the complete folder, including references. Preserve local changes and verify loading again.
$editorial-studio Create a photography-led architecture onepager for design readers. Start with the purpose, audience and style. Ask for an inspiration URL or no example, then show two distinct creative directions before production.Expect a conversation before a layout. You approve the direction and exact images. The final PDF still needs a visual review.
Photography is researched and reviewed by a person. Provenance and rights evidence are recorded; approval fields do not prove a licence. Missing or unapproved images block export.
Release 0.4.0 passed 41 regression tests on Windows and Linux, with Python 3.11 and 3.13. A dependency scan found no known vulnerabilities at review time. This is not a security certification.
RELEASE REVIEW / 6 NOVEMBER 2026