How do I install the Playwright Codex skill?

Run npx skillmds@latest add cyz26/playwright-codex in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is Playwright Codex safe to use?

This skill has not completed SkillMD's automated safety review yet. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with Playwright Codex?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is Playwright Codex free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published Playwright Codex?

cYz26 (@cyz26) published this skill. Their other Agent Skills are listed on their SkillMD profile.

Playwright Codex by cyz26 — Agent Skill

Name: Playwright Codex
Author: cYz26

File contents

Playwright Codex Skill

Use this skill when the task is about configuring Playwright for Codex, deciding between CLI and MCP, or running browser automation in a Codex-friendly workflow.

This skill is a Codex-specific companion to the upstream playwright skill. It does not replace $playwright for ordinary browser interaction tasks.

Default mode

Use Playwright in Codex with a CLI-first workflow:

/Users/cY/.codex/bin/pwcli-codex open https://playwright.dev --headed
/Users/cY/.codex/bin/pwcli-codex snapshot
/Users/cY/.codex/bin/pwcli-codex screenshot

Keep MCP enabled as a fallback for tasks that need tool-native persistent context or richer page introspection inside Codex.

Keep CLI session names short. Overly long --session values can exceed the local socket path limit used by playwright-cli.

Decision tree

If the task is regular browser automation during coding work, use the CLI launcher.
If the task needs a persistent browser tool loop inside Codex, use MCP.
Do not default to @playwright/test or new test specs unless the user explicitly asks for tests.

Headed vs headless

Default to --headed for:

initial UI debugging
visual verification
animation, hover, focus, overlay, scrolling, drag, and responsive issues

Use headless for:

smoke checks
fixed command sequences
console, network, cookie, or storage inspection
screenshot, pdf, trace, or video capture where a live window is not needed

Headless can still verify UI through screenshots and traces, but it does not provide a live window. When a headless result is unclear, re-run it in --headed mode before adding custom scripting.

Artifact policy

Prefer output/playwright/<label>/ when the repository already uses that convention.
Otherwise use /tmp/codex-playwright/<label>/.
Always capture artifacts when they help with debugging: screenshot, console, network, and tracing.

Snapshot rules

Always run snapshot before using element refs.

Run snapshot again after:

navigation
clicks that change the UI
modal or menu open and close
tab switches
any stale ref error

References

Open only what you need:

Codex playbook: references/codex-playbook.md
Upstream browser CLI skill: /Users/cY/.codex/skills/playwright/SKILL.md

---name: playwright-codexdescription: Configure and use Playwright in Codex with a CLI-first workflow and MCP fallback.---# Playwright Codex SkillUse this skill when the task is about configuring Playwright for Codex, deciding between CLI and MCP, or running browser automation in a Codex-friendly workflow.This skill is a Codex-specific companion to the upstream `playwright` skill. It does not replace `$playwright` for ordinary browser interaction tasks.## Default modeUse Playwright in Codex with a CLI-first workflow:```bash/Users/cY/.codex/bin/pwcli-codex open https://playwright.dev --headed/Users/cY/.codex/bin/pwcli-codex snapshot/Users/cY/.codex/bin/pwcli-codex screenshot```Keep MCP enabled as a fallback for tasks that need tool-native persistent context or richer page introspection inside Codex.Keep CLI session names short. Overly long `--session` values can exceed the local socket path limit used by `playwright-cli`.## Decision tree1. If the task is regular browser automation during coding work, use the CLI launcher.2. If the task needs a persistent browser tool loop inside Codex, use MCP.3. Do not default to `@playwright/test` or new test specs unless the user explicitly asks for tests.## Headed vs headlessDefault to `--headed` for:- initial UI debugging- visual verification- animation, hover, focus, overlay, scrolling, drag, and responsive issuesUse headless for:- smoke checks- fixed command sequences- console, network, cookie, or storage inspection- screenshot, pdf, trace, or video capture where a live window is not neededHeadless can still verify UI through screenshots and traces, but it does not provide a live window. When a headless result is unclear, re-run it in `--headed` mode before adding custom scripting.## Artifact policy- Prefer `output/playwright/<label>/` when the repository already uses that convention.- Otherwise use `/tmp/codex-playwright/<label>/`.- Always capture artifacts when they help with debugging: `screenshot`, `console`, `network`, and `tracing`.## Snapshot rulesAlways run `snapshot` before using element refs.Run `snapshot` again after:- navigation- clicks that change the UI- modal or menu open and close- tab switches- any stale ref error## ReferencesOpen only what you need:- Codex playbook: `references/codex-playbook.md`- Upstream browser CLI skill: `/Users/cY/.codex/skills/playwright/SKILL.md`

Frequently asked questions

Playwright Codex