cep-test-authoring

Converts plain-language intent files into deterministic Playwright tests for scheduled check_cep monitoring.

2|Updated Mar 2, 2026
One-click install
npx skills add https://github.com/lausser/check_cep --skill cep-test-authoring-lausser
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: cep-test-authoring
Source: https://github.com/lausser/check_cep/tree/main/.agents/skills/cep-test-authoring
Command: npx skills add https://github.com/lausser/check_cep --skill cep-test-authoring-lausser

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Non-technical users can describe a web flow in plain language, but turning that description into a reliable, deterministic Playwright check that runs on a schedule without any AI at run time is hard. This Skill guides the full authoring session from intent file to packaged, deployable test. ## Core Features & Use Cases - Structured authoring workflow: Walks through PREFLIGHT, EXPLORE, ASSESS, AUTHOR, RUN, and COMPARE phases, validating the intent file and confirming stable Playwright locators before writing any test code. - Quality gate before packaging: Enforces a nine-item checklist including consecutive clean runs, an independent critic pass, perfdata emission, and a VERIFICATION.md report before the test can be packaged with cep_package. - Site knowledge accumulation: Reads and updates a site-level KNOWLEDGE.md so consent dialogs, login quirks, and reliable selectors are reused across sessions instead of rediscovered. - Use Case: An admin runs check_cep --authoring with an intent file describing a login-and-check-balance flow; the Skill explores the site, authors the test reusing sanctioned robustness helpers, proves it green across consecutive runs, and produces a deployable tar.gz package. ## Quick Start Start an authoring session with check_cep --authoring and ask the assistant to turn your intent file into a deployable Playwright check.

Frequently Asked Questions about cep-test-authoring

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I turn a plain-language intent file into a Playwright test?

Write an intent file with Target URL, Desired Flow, Success Criteria, and Lessons Learned sections, then start check_cep --authoring. The workflow explores the site, confirms stable locators, authors the test, and runs it until consecutively green before packaging.

How do I find stable Playwright locators for a web check?

Discover elements with chrome-devtools-mcp by inspecting the DOM, network, and console, then confirm each candidate as a Playwright locator preferring getByRole, getByText, getByLabel, or getByTestId. A good locator resolves to exactly one element.

When should I use vision template matching instead of DOM locators?

Use check-cep-vision template matching only when no stable DOM locator exists, such as canvas, WebGL, dynamic class names, or PDF content. DOM locators are always preferred first; user-attached screenshots can serve as candidate templates.

Where do credentials go in a check_cep test?

Credentials belong only in a .env file loaded through a helper, never inline in the test, helpers, README, KNOWLEDGE.md, VERIFICATION.md, or the intent file. The packaging tool refuses to build if it finds a secret outside .env.

Why does a single green Playwright run not qualify a test for deployment?

Flakiness typically surfaces on the second or third run, so the quality gate requires consecutive clean runs (default 2, or 3 for volatile sites). An independent critic pass must also return pass before packaging.

Can the authored test use AI or thresholds at run time?

No. Deployed checks run deterministically with no AI, no credentials beyond .env, and no browser-inspection tooling. Threshold wishes are handled downstream by the alerting system; the test only emits measurements as perfdata.