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.
The body is a well-structured, highly actionable guide: executable commands, an ordered workflow with a mandatory decision gate, and clean progressive disclosure into verified reference files. Its main weakness is mild redundancy (repeated naming rules, overlap between the decision gate and routing table) and a missing explicit validate→retry loop.
Suggestions
Consolidate the app-name rule (≤26 chars, lowercase letters/numbers/hyphens) into a single canonical location and reference it, instead of restating it in Generic Guidelines, the Scaffolding step, and Directory Naming.
Add an explicit validate→fix→re-validate feedback loop to the Analytics development workflow (e.g. 'If validate fails, review errors, fix, and re-run validate before proceeding').
Reconcile the Data Access Decision Gate table and the 'When to Use What' routing list so each pattern is defined once and the other cross-references it, reducing overlap.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Information-dense with terse bullets and tables, and it assumes Claude's competence (no padding about what Databricks or React are). Minor redundancy holds it below 5: the ≤26-char lowercase-hyphen app-name rule is repeated in three places, and the Decision Gate table overlaps with the 'When to Use What' routing list. | 4 / 5 |
Actionability | Provides copy-paste-ready, executable commands throughout ('databricks apps init --name … --features … --set …', 'npm run typegen', 'databricks apps validate --profile <PROFILE>', 'npx @databricks/appkit docs') plus a wrong-vs-right scaffolding example covering the common mistake. | 5 / 5 |
Workflow Clarity | Clear ordered sequencing ('FOLLOW THIS ORDER'), a mandatory Data Access Decision Gate checkpoint, numbered analytics steps 1–6 with a typegen 'verify all queries show ✓' checkpoint, validate-before-deploy, and post-deploy verification. Held below 5 by the absence of an explicit validate→fix→re-validate feedback loop in the main analytics workflow. | 4 / 5 |
Progressive Disclosure | SKILL.md is an overview pointing to ~16 real one-level-deep reference files via a 'Required Reading by Phase' table and 'When to Use What' routing; content is appropriately split with easy navigation and no nested reference chains. | 5 / 5 |
Total | 18 / 20 Passed |