Content
63%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 operational guide: concrete commands, copy-paste SQL examples, explicit decision rules, and real validation checkpoints (dry-run, lock diffing, preview-before-deploy). Its weaknesses are repetition and dense prose that could be tightened, a missing example for the most common MySQL case (declaring the database connection as a resource parameter), and substantial generic platform content inlined where a shared reference would keep the MySQL skill lean.
Suggestions
State the 'it is not a deploy' rule once (e.g. in the command list) and remove the repetitions in the metadata-sync and after-writing sections; split long parenthetical-heavy paragraphs into short imperative bullets.
Add a concrete MySQL example showing how a script declares its database connection (resource-typed parameter syntax for SQL) and a worked 'wmill script preview <path> -d ...' invocation with real placeholder args.
Move the language-agnostic Windmill content (General Principles, Return Values, Preprocessor Scripts) into a shared reference file and keep only the MySQL-specific guidance plus a clearly signaled pointer in SKILL.md.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with non-obvious operational guidance rather than known-concept padding, but it is noticeably repetitive and could be tightened: 'not a deploy' is restated three times (lines 14, 31, and 45), and several long run-on paragraphs (e.g. the metadata-sync and resource-type guidance) bury key rules in qualifiers. It sits above anchor 2 because most content earns its tokens, and below anchor 4 because the repetition and parenthetical-heavy prose are real trimming opportunities. | 3 / 5 |
Actionability | Guidance is mostly executable: concrete wmill commands with flags ('wmill script preview <script_path>', 'generate-metadata --dry-run', 'rehash', 'resource-type list --schema') and copy-paste-ready SQL examples for parameter naming, JSON_TABLE/S3Object consumption, and the '-- s3' streaming directive. It falls short of anchor 5 because the most common case — how a MySQL script declares its database connection/resource-typed parameter — is never shown in the SQL section, and the preview invocation 'wmill script preview <path> -d '<args>'' lacks a concrete worked example. | 4 / 5 |
Workflow Clarity | The workflow (write → preview locally → generate-metadata after edits → deploy only on explicit request) is clearly sequenced with decision rules for preview vs run vs sync push, and genuine checkpoints are present: dry-run to inspect stale items before regenerating, diffing the regenerated locks and reporting version changes, preview validating before any deploy, and confirming before side-effecting execution. Because validation is present, the database-operations cap at 3 does not apply; it misses anchor 5 mainly because the end-to-end flow must be assembled from separate sections rather than one explicit sequence. | 4 / 5 |
Progressive Disclosure | Sections are well-organized with clear headers, but the skill is over 50 lines and inlines roughly 40 lines of generic Windmill platform content (General Principles, Return Values, Preprocessor Scripts, multi-language entrypoint details) that belongs in a shared reference file for a MySQL-specific skill. The only external pointers ('AGENTS.wmill.md', the 'preview' skill) are to files outside the bundle. This matches anchor 3 — 'content that should be separate is inline' — rather than anchor 4's 'most content is appropriately placed'. | 3 / 5 |
Total | 14 / 20 Passed |