Content
68%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-structured, highly actionable workflow skill with concrete commands, verified script options, and a clearly sequenced 7-step process that appropriately offloads title formulas to a real one-level reference. Its main weaknesses are the missing safety/validation checkpoint around the destructive in-place typography path (capping workflow clarity) and localized verbosity in runtime-mechanics sections.
Suggestions
Add a backup step before the in-place "Typography only" path (Step 1 option 3) — e.g., copy the original file with a timestamped suffix, mirroring the Step 5 backup — so the destructive modification is reversible.
Add an explicit validation loop after Step 6: check the typography script's exit status/output, and on failure review the error and re-run before reporting completion, rather than reporting applied/skipped unconditionally.
Trim the "User Input Tools" section to a single line (e.g., "Prefer the runtime's built-in user-input tool; otherwise ask numbered plain-text questions") and drop the duplicated EXTEND.md mention in "Extension Support".
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and table-driven (preferences tables, options table, do/don't lists) with almost no concept explanation Claude already knows, but sections like the ~10-line "User Input Tools" runtime mechanics and the duplicated EXTEND.md pointer are over-explanation that could be trimmed — efficient with minor instances of padding, anchor 4; not 3 because the padding is localized rather than pervasive. | 4 / 5 |
Actionability | Guidance is mostly executable: a copy-paste bash backup snippet, a script invocation `${BUN_X} {baseDir}/scripts/main.ts {output-file-path} [options]` whose flags were verified against the actual scripts/main.ts CLI, exact EXTEND.md paths, and verbatim prompt/report templates. Minor gaps remain — commands require placeholder substitution and the usage examples line ("main.ts article.md (defaults)") omits the full invocation prefix — so anchor 4 rather than the fully copy-paste-ready anchor 5. | 4 / 5 |
Workflow Clarity | Steps 1-7 are clearly sequenced with explicit decision branches at Step 1 ("Optimize → Step 2", "Keep original → skip to Step 5", "Typography only → skip to Step 6"), but the "Typography only" path runs the script on the original file in-place — a destructive modification with no backup step (the only backup covers the -formatted.md copy) and no verification of script success before reporting. Per the rubric, a destructive operation without validation/feedback loops caps this dimension at 3, taking precedence over the otherwise clear sequence. | 3 / 5 |
Progressive Disclosure | The single external reference [references/title-formulas.md](references/title-formulas.md) is well-signaled, one level deep, and verified to exist; the three scripts are described in a table rather than inlined. However, the ~236-line body carries several large verbatim prompt/report templates (Step 1 menu, Step 2 analysis template, Step 7 report) that make it denser than an overview — good structure with minor organization gaps, anchor 4; not 5 because not all content is appropriately split out of the main file. | 4 / 5 |
Total | 15 / 20 Passed |