Content
86%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-crafted, highly actionable skill: concrete commands, exact paths, sensible two-branch workflow, and a genuinely useful set of project-specific translation rules. The main weaknesses are the absent locale guideline files (broken reference structure in this bundle) and the lack of an explicit fix-and-re-run loop in the Validation section.
Suggestions
Ship (or verify) the 21 locales/*.md guideline files referenced in the Locale-Specific Guidelines table — as provided, every reference in the table is broken and the skill's core glossary/tone content is unreachable.
Add an explicit feedback loop to the Validation section, e.g. 'If the completeness check reports missing/TODO/unused entries, fix them and re-run until the table shows 0 missing, 0 TODOs, 0 unused' and 'If prek reports errors, fix them and re-run the affected hooks.'
Briefly state the precondition for running breeze/prek (dev environment setup, working directory at the repo root) so the commands are executable on the first try.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and assumes Claude's competence: 'Replace every `TODO: translate: <English terminology>` entry, including the prefix, with the translated string' and 'breeze ui check-translation-completeness --language <locale> --add-missing' give pure project-specific instruction with no padding or explanation of concepts Claude already knows. Every section (terms-kept-in-English table, plural suffixes, hotkeys, variables) carries non-obvious Airflow-specific rules that earn their tokens. | 5 / 5 |
Actionability | Guidance is fully executable: exact paths ('airflow-core/src/airflow/ui/public/i18n/locales/<locale>/'), copy-paste breeze and prek commands with flags, concrete config snippets for each of the three files to update, and an example of the generated scaffold JSON. The <locale>/<suffixes> placeholders are task parameters rather than pseudocode, and expected validation output ('0 missing, 0 TODOs, 0 unused, and 100% coverage') is specified. | 5 / 5 |
Workflow Clarity | Clear two-branch dispatch ('If it does, skip ahead to Updating an Existing Translation. If not, start with Adding a Translation'), well-sequenced steps in both branches, and a Validation section with explicit expected outcomes. Not a 5 because there is no explicit error-recovery loop after validation — the text says what the output should show but never instructs to fix and re-run if the completeness check or prek hooks fail. | 4 / 5 |
Progressive Disclosure | The locale-guideline table is well-designed — 21 one-level-deep, clearly signaled references — but none of the referenced locales/*.md files exist in the provided bundle (references/, scripts/, assets/ are absent entirely), so every guideline link is unresolvable as shipped here. Good in-body structure, but the content split cannot be verified, which is more than the 'minor organization gaps' of a 4. | 3 / 5 |
Total | 17 / 20 Passed |