Content
78%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 strong, executable CLI reference: concrete commands, explicit anti-patterns, troubleshooting, and gated setup steps with real checkpoints. Its two weaknesses are a deploy workflow that never explicitly sequences validation before deploy, and a progressive-disclosure layer whose four referenced detail files do not exist in the bundle, leaving the deepest guidance unreachable. Conciseness is good with only minor duplication between the reading-table and reference-list sections.
Suggestions
Add the four referenced files (databricks-cli-install.md, databricks-cli-auth.md, data-exploration.md, declarative-automation-bundles.md) to the bundle, or move their content inline — currently every 'Read BEFORE proceeding' link points to a file that doesn't exist, breaking the skill's own gating rules.
Make the deploy sequence explicit with a validation gate: '1. databricks bundle validate --profile <PROFILE> 2. Only when valid: databricks bundle deploy -t <TARGET> --profile <PROFILE>' in the Quick Reference's bundles section.
Merge the 'Reference Guides' bullet list into the 'Required Reading by Task' table to remove the duplicate listing of the same four files and free tokens for the missing deploy-validation step.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is a dense command reference with no concept re-explanations — it assumes Claude's competence throughout ('NEVER auto-select a profile', works/doesn't-work shell patterns). Not score 5 because of mild redundancy: the four reference files are listed twice (in 'Required Reading by Task' and again in 'Reference Guides'), and profile-selection guidance repeats the prerequisites section's auth profiles step. Not score 3 because every section carries non-obvious, task-critical information. | 4 / 5 |
Actionability | Fully executable, copy-paste-ready commands with explicit placeholders for every common case: listing, getting details, UC positional-argument examples, bundle lifecycle, and AI-tools queries. It goes beyond happy-path instructions by showing anti-examples ('❌ WRONG: these flags/commands DON'T EXIST'), a troubleshooting table mapping real error strings to fixes, and a 'check --help when in doubt' fallback. This matches the anchor for executable commands covering the common cases. | 5 / 5 |
Workflow Clarity | The setup workflow has explicit checkpoints: version check with a hard STOP gate ('If the CLI is missing or outdated... STOP'), auth verification, and a gated 4-step profile-selection sequence that requires user confirmation. Not score 5 because deploy ('databricks bundle deploy') lacks an explicit validate-before-deploy checkpoint in sequence — 'bundle validate' appears only as a bare command in the Quick Reference, and the guidance for deploy-time validation is delegated to a reference file. Not score 3 because validation gates are present and explicit for the risky setup steps. | 4 / 5 |
Progressive Disclosure | The structure itself is good — a task-routed 'Required Reading by Task' table plus one-level-deep markdown links — but all four referenced files (databricks-cli-install.md, databricks-cli-auth.md, data-exploration.md, declarative-automation-bundles.md) are absent from the bundle (no references/ directory or matching files exist anywhere), so every deep-dive link dead-ends and the disclosure chain is broken. Not score 4 because navigation to detail content actually fails; not score 2 because the in-body overview is well organized and the reference intent is clearly signaled. | 3 / 5 |
Total | 16 / 20 Passed |