Playwright Interactive Debugger
Iterate on local web and Electron UI debugging with persistent browser sessions.
The documentation clearly discloses the js_repl, project-directory, and Playwright prerequisites, and describes persistent sessions, screenshots, and Electron main-process inspection. However, it requires disabling the sandbox with danger-full-access, leaves npm install playwright unpinned, and does not establish sensitive-data isolation, user confirmation, least privilege, or rollback, so substantial points are deducted.
The workflow, Web/Electron branches, reload and restart rules, stale-handle handling, and some failure messages are reasonably explicit and statically reproducible as instructions. However, there are no executed tests, several paths are explicit happy paths, and coverage of dependency failures, server-start failures, abnormal input, and cleanup failures is thin. Static calibration caps this below 10.
The audience and scenarios are clear: local Web and Electron debugging, functional QA, visual QA, responsive viewports, and persistent sessions. Boundaries and trigger conditions are not systematically stated, and use depends on js_repl, specific Codex configuration, and elevated sandbox settings. The core function targets local environments rather than unreachable overseas services, so no additional mainland-network deduction is applied.
The skill is well organized, with prerequisites, installation, session modes, checklists, screenshot examples, and limitations. However, the selected skill's license, version, changelog, maintenance responsibility, and update path are not clear from the skill evidence. The shared README also says the repository is deprecated and points elsewhere, so governance and lifecycle information are incomplete.
The skill provides directly usable patterns for persistent Playwright debugging, functional QA, and visual QA, and its core value proposition is clear. Nevertheless, representative outputs and execution evidence are absent, while configuration requirements, elevated permissions, and the extensive manual QA process increase cost and uncertainty. Static evidence does not justify a higher effectiveness score.
The source includes many auditable commands, code examples, and QA criteria, making the intended procedure reviewable. It lacks a skill-specific test suite, CI coverage, or third-party execution evidence, so the assessment rests mainly on one documentation source. Static calibration caps this at 5.
- Do not follow the default sandbox-disabling guidance in a workspace containing credentials, production data, or sensitive user information; confirm permission boundaries and isolate test data first.
- npm install playwright is unpinned, so dependency provenance, browser-binary installation, and supply-chain risk require separate review.
- This assessment did not execute js_repl, Playwright, a Web app, or Electron; the documented key paths are not proven runnable.
- The repository README marks the repository as deprecated; confirm the skill's current maintenance source and license before adoption.
What it does & when to use it
This skill uses a persistent Playwright session through `js_repl` to debug local web and Electron applications. It keeps browser, page, context, and application-window handles alive across iterations. The workflow covers functional QA, separate visual QA, viewport-fit checks, and screenshot capture. It is a strong fit for iterative UI debugging, provided the Codex session and local Playwright environment are configured as required.
Initializes an npm project when needed and installs Playwright; launches Chromium or Electron through js_repl; opens local applications while reusing browser, page, context, or Electron-window handles; performs functional QA with keyboard, mouse, touch, and equivalent Playwright inputs; runs separate visual QA, viewport-fit checks, and screenshot capture; reloads renderer-only changes and relaunches Electron after main-process, preload, or startup changes.
- A frontend developer needs to repeatedly verify interaction flows and visual states in a local web app.
- An Electron developer changes main-process, preload, or startup code and needs a fresh process for validation.
- A UI debugger checks clipping, overflow, and layout behavior across desktop and mobile viewports.
- A QA engineer needs real user-input coverage for visible controls, reversible states, and a critical end-to-end flow.
- A local-app team needs screenshots as evidence for visual QA signoff.
Pros & cons
- Persistent Playwright handles support fast iterative debugging.
- The workflow covers both web and Electron applications.
- Functional QA, visual QA, and viewport-fit validation are explicitly separated.
- Desktop, mobile, native-window, and screenshot-normalization procedures are documented.
- It requires Codex `js_repl` and a new Codex session after enabling it.
- The current workflow requires disabling sandboxing with `danger-full-access`, which is a significant permission requirement.
- Node.js, npm, and Playwright must be installed locally; Chromium or Electron may also be needed.
- The source provides no test suite, platform coverage matrix, or empirical test results.
- The repository README says the repository is deprecated and points readers to the OpenAI Plugins repository.
How to install
Install the curated skill in Codex with $skill-installer playwright-interactive, then restart Codex. In the target project directory, run test -f package.json || npm init -y, npm install playwright, and the documented Node import check: node -e "import('playwright').then(() => console.log('playwright import ok')).catch((error) => { console.error(error); process.exit(1); })". For headed Chromium web testing, optionally run npx playwright install chromium; install Electron when the Electron workflow requires it.
How to use
Enable js_repl; if it is missing, add [features] and js_repl = true to ~/.codex/config.toml, then start a new Codex session. Run the required development server from the target project directory, launch the local web or Electron app in a persistent js_repl session, and reuse the Playwright handles. Build and maintain a QA inventory covering requirements, controls, states, and final claims; perform functional QA and a separate visual QA pass; clean up the Playwright session when finished. The SKILL.md does not define a fixed user trigger phrase.