engineering / skill

Use Both: Claude + Codex Workflow Kit

Six Claude and Codex workflows with prompts, templates and a guide, plus handoff and prime skills for moving a project between assistants.

Repository owner promptadvisers

Not rated yet
0 comments

Like 0 Save

Published 11 Oct 2026Listing updated 12 Oct 2026

At a glance

Purpose: Six Claude and Codex workflows with prompts, templates and a guide, plus handoff and prime skills for moving a project between assistants.
Runs in: Local agent with shell
Operating systems: Not stated
Needs: Runtime: Text editor for the no-install route; Needs internet: Not stated; Needs accounts or API keys beyond your agent: Not stated
Files and data: Reads files: Yes; Writes files: Yes; Sends data to other services: Not stated
Tool type: Skill
Delivery: From GitHub
Version: Latest on the main branch
Access: Code on GitHub
Compatibility: According to the README: Claude Code and Codex. In Claude Code the skills run as /handoff and /prime. In Codex they run as $handoff and $prime, where skills are available. With any other assistant, the README says to ask it to read handoff/LATEST.md and the referenced snapshot. The installer and example need Python 3.9 or newer, and the branch and review workflow needs Git.
External costs: The README says the public files are free without a purchase. It says a Gumroad edition packages the same kit and PDF together, with any contribution optional, and that your AI subscriptions, API usage and third-party services are separate. The repository is MIT-licensed.
Permissions: The skills act through your coding agent with the access you give it. The README describes a preview-first installer with project-local destinations under .claude/skills and .agents/skills and refusal to overwrite existing skills. The README describes handoff as writing a project snapshot and prime as reading stored state. The README also includes branch-based building and review, image generation and supervised computer use, with credentials, payment and consequential submissions kept under the user's control. The README says a skill is an instruction file rather than an enforcement boundary.
Usage terms: MIT License, as stated in the repository. See its LICENSE file for the exact terms. The README says the MIT License covers the original text and code, and that third-party product names and marks remain their owners' property, with details in ASSET-NOTICE.md.

Code on GitHub

promptadvisers/claude-codex-workflow-kit

MIT

Python18 stars3 forksLast pushed 26 Sept 2026 UTC

Use Both: six Claude + Codex workflows, nine prompts, public-safe prime and handoff skills, and a printable guide.

Public GitHub data fetched 13 Oct 2026 UTC · 0 days old

Skill folders in this repository (2)

Folders that contain a SKILL.md file, from GitHub's file list on 13 Oct 2026 UTC.

View on GitHub ↗

What it does

Use Both is a kit for working with Claude and Codex on one project. According to the README, it is a companion to a video by Mark Kashef. The README lists six workflows: let one assistant critique another's draft, generate images, split building and review, supervised computer use, give an assistant a finish line, and hand work off between assistants.

The repository holds a step-by-step guide, copy-ready prompts, a printable PDF guide, templates for a task brief, plan, review and handoff, and a fictional worked example. It also holds notes on setup, privacy and safety, and troubleshooting.

The README lists handoff and prime skills. The README describes handoff as a project-local skill that writes a compact snapshot, and prime as a read-first skill that recovers state without treating old notes as new permission. The README describes a preview-first installer script that writes them into a project for Claude Code, Codex or both.

Who it is for

People who use two AI assistants, such as Claude and Codex, and want workflows and prompts for planning, review and handoff between them.

Example tasks

Follow the five-minute, no-install trial in START-HERE.md.
Pick a task from PROMPTS.md, replace the bracketed placeholders and run it in your own project.
Have one assistant build on a branch and the other review the diff.
Run the installer in preview mode to see where the handoff and prime skills would be written.
In Claude Code, run /handoff at the end of a session and /prime in the next one.

Getting started

The README offers a no-install route: read START-HERE.md and select a task from PROMPTS.md. For skill installation, the README shows a preview command targeting an existing project you own, followed by a command with --apply after inspecting the destinations. The README gives claude, codex and both targets and names project-local destinations under .claude/skills/{handoff,prime}/SKILL.md and .agents/skills/{handoff,prime}/SKILL.md. The README describes refusal to overwrite existing skills and points to docs/setup.md for manual installation. The README has the exact commands.

Limitations

According to the README, the no-install route needs two assistants and a text editor. CLI delegation needs installed, authenticated Claude Code and Codex CLIs and a terminal-capable host. Access to particular models and tools depends on your account and environment. The installer and example use Python 3.9 or newer with no third-party Python packages. Git is needed for the branch and review workflow. The README says image generation and computer use are not guaranteed by having the CLI alone. It says the video names Claude Opus 5.5 and GPT-6 Astra, and that these are not a promise that your account exposes the same models, effort settings, prices or tools. The README says its tests do not prove that every model follows the skills. It says its documentation was checked on September 26, 2026, and advises preferring current first-party documentation when interfaces change. It calls itself an independent educational resource, not an OpenAI or Anthropic product. The repository can change at any time; check its README before you install.

Data practices

The README points to docs/privacy-and-safety.md for data boundaries, approvals, costs and handoffs. It says the skills do not include private business context, client information or credentials, and that you should never post credentials or private handoffs in GitHub issues. Where data goes otherwise depends on your coding agents and model providers.