Content
77%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
A well-structured, actionable workflow: concrete locator and assertion rules, an explicit run-and-fix verification loop, and convention-matching guidance. The main defects are the unverifiable, inconsistently-prefixed template references that undermine navigation, and a placeholder-only main code example where one complete generated test would set the quality bar.
Suggestions
Fix the template references: use one consistent path style, and align the table with the actual bundle layout (e.g. 'templates/auth/login.md' for all rows) so the referenced files resolve.
Replace or supplement the placeholder skeleton in step 4 with one complete, runnable generated example test so the expected output quality is unambiguous.
Trim filler comments in the structure block ('// Group related behaviors', '// Arrange/Act/Assert' narration) and fold the overlapping 'Bare CSS selectors' item in the Never-use list into the locator priority.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes competence: no 'what is Playwright' padding, and the locator-priority list, GOOD/BAD assertion contrast, and Never/Always rules each earn their tokens. Minor trims remain (skeleton filler comments like '// Group related behaviors', and the 'Never use' list partially overlapping the locator priority), so it sits just below the 'every token earns its place' anchor. | 4 / 5 |
Actionability | Highly concrete guidance: exact locator method priorities, an executable GOOD/BAD assertion pair, an exact verify command ('npx playwright test <generated-file> --reporter=list'), and convention-matching rules. The primary structure example is a placeholder skeleton ('test(\u2019should <expected behavior>\u2019...)') rather than complete runnable code, which is a minor gap against the fully copy-paste-ready anchor. | 4 / 5 |
Workflow Clarity | Seven clearly sequenced steps (understand target, explore codebase, select templates, generate, match conventions, supporting files, verify) culminate in an explicit validation step with a feedback loop: read error, fix the test, re-run, and escalate app issues to the user. This matches the anchor for clear sequence with explicit validation and error-recovery loops. | 5 / 5 |
Progressive Disclosure | The single file is well-sectioned and template references are surfaced in a clear table, but the referenced paths are inconsistent ('../pw/templates/auth/login.md' vs 'templates/crud/', 'templates/checkout/', ...) and no templates/ directory exists in the bundle, so the navigation targets are unverifiable and the path prefix is unreliable. | 3 / 5 |
Total | 16 / 20 Passed |