Name: generate-tests
Author: saikumarreddyk510-byte

Prompt file imported from saikumarreddyk510-byte/plywrghtTmpt (.github/prompts/generate-tests.prompt.md). Fill in {{task}} before use. Copyright stays with the author.

Follow the process in .claude/skills/generate-tests/SKILL.md exactly. It is the single source of truth for this step — used by both the Claude Code and Copilot pipelines. Read it, .claude/skills/playwright-best-practices/SKILL.md, .claude/skills/app-domain/SKILL.md, and docs/pipeline/test-strategy.md before writing anything.

Copilot-specific mechanics:

Real-browser verification uses the Playwright MCP server already registered in .vscode/mcp.json (the same server Claude Code's pipeline uses via .mcp.json) — use it to confirm selectors against the live page before writing the Page Object, and again while debugging a failure.
Run tests from the integrated terminal: npx playwright test <spec>.spec.ts --headed --project=<name>. Do not stop after writing — the test is only done when it passes for real, in a real browser.
Locator-class failures (timeout / not found / strict-mode violation) during the debug loop → follow /heal-test's procedure instead of guessing.

Feature/flow to implement: {{task}}

Follow the process in `.claude/skills/generate-tests/SKILL.md` exactly. It is the single source of truth for this step — used by both the Claude Code and Copilot pipelines. Read it, `.claude/skills/playwright-best-practices/SKILL.md`, `.claude/skills/app-domain/SKILL.md`, and `docs/pipeline/test-strategy.md` before writing anything. Copilot-specific mechanics: - Real-browser verification uses the Playwright MCP server already registered in `.vscode/mcp.json` (the same server Claude Code's pipeline uses via `.mcp.json`) — use it to confirm selectors against the live page before writing the Page Object, and again while debugging a failure. - Run tests from the integrated terminal: `npx playwright test <spec>.spec.ts --headed --project=<name>`. Do not stop after writing — the test is only done when it passes for real, in a real browser. - Locator-class failures (timeout / not found / strict-mode violation) during the debug loop → follow `/heal-test`'s procedure instead of guessing. Feature/flow to implement: {{task}}

{ "prompt": "Follow the process in `.claude/skills/generate-tests/SKILL.md` exactly. It is\nthe single source of truth for this step — used by both the Claude Code and\nCopilot pipelines. Read it, `.claude/skills/playwright-best-practices/SKILL.md`,\n`.claude/skills/app-domain/SKILL.md`, and `docs/pipeline/test-strategy.md` before\nwriting anything.\n\nCopilot-specific mechanics:\n- Real-browser verification uses the Playwright MCP server already registered\n in `.vscode/mcp.json` (the same server Claude Code's pipeline uses via\n `.mcp.json`) — use it to confirm selectors against the live page before\n writing the Page Object, and again while debugging a failure.\n- Run tests from the integrated terminal: `npx playwright test <spec>.spec.ts\n --headed --project=<name>`. Do not stop after writing — the test is only\n done when it passes for real, in a real browser.\n- Locator-class failures (timeout / not found / strict-mode violation) during\n the debug loop → follow `/heal-test`'s procedure instead of guessing.\n\nFeature/flow to implement:\n{{task}}", "variables": [ "task" ], "notes": "Write Playwright TypeScript E2E tests with real-browser validation and a self-healing debug loop (mirrors the generate-tests Claude Code skill). Mode: agent.", "metadata": { "source": { "provider": "github-prompts", "repository": "https://github.com/saikumarreddyk510-byte/plywrghtTmpt", "path": ".github/prompts/generate-tests.prompt.md", "ref": "de10f87829ebfbc5ec87d9788c6c36661c171a1a", "url": "https://github.com/saikumarreddyk510-byte/plywrghtTmpt/blob/de10f87829ebfbc5ec87d9788c6c36661c171a1a/.github/prompts/generate-tests.prompt.md", "key": "saikumarreddyk510-byte/plywrghtTmpt/.github/prompts/generate-tests.prompt.md" } } }

generate-tests

About

Use it

Manifest

Reviews