Content
81%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.
The body delivers an exceptionally clear, enforced workflow with executable commands, concrete error-code mappings, and strong validation feedback loops, backed by a well-structured one-level-deep reference bundle. The main weakness is redundancy: repeated dbt1065 guidance and a recap-heavy Important Notes section inflate token cost without adding new information.
Suggestions
Remove the duplicated dbt1065 'ignore these warnings' paragraph — it appears nearly verbatim in both 'Hard rules' and 'Important Notes'; state it once.
Cut or drastically trim the 'Important Notes' section: most bullets restate rules already covered in 'Mandatory execution order', 'Step 1', and 'Progressive Fixing Approach'.
Move the full Category A/B pattern lists and the 'Presenting Findings' / 'Track progress' templates into references/ (e.g. classification-categories.md or a new presentation.md), keeping only a summary and links in SKILL.md to tighten progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient, but the dbt1065 'ignore these warnings' guidance appears twice nearly verbatim (Hard rules and Important Notes), and the Important Notes section restates rules already given in the workflow sections (autofix first, review diffs, validate after each fix, suggest dbt debug). This is more than minor trimmable material, so it sits at the 'mostly efficient but could be tightened' anchor rather than 4. | 3 / 5 |
Actionability | Fully executable guidance throughout: copy-paste commands (dbt debug, uvx --from git+...dbt-autofix deprecations, git stash create), exact error-code-to-fix mappings (e.g. dbt1501: config.require('meta').key to config.meta_require('key')), a concrete WebFetch GitHub-search URL template, and ready-to-use output format templates. Common cases are covered. | 5 / 5 |
Workflow Clarity | Mandatory numbered execution order with hard rules, a violation-recovery protocol ('acknowledge the violation... execute that step now'), and an explicit validate-fix-retry feedback loop ('After EVERY fix, re-run the repro command'), plus handling for cascading errors and same-node latent duplicates. This matches the anchor-5 checklist pattern. | 5 / 5 |
Progressive Disclosure | Three reference files are well signaled, one level deep, indexed by a README, and every in-body anchor (case-sensitive-identifier-mismatches, verifying-against-the-real-warehouse..., guardrail-...) resolves to a real heading. Kept at 4 rather than 5 because the body retains full category pattern lists and presentation templates inline and duplicates guidance across sections — a minor organization gap rather than a clean overview-plus-references split. | 4 / 5 |
Total | 17 / 20 Passed |