Content
85%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.
An exceptionally actionable, well-sequenced skill with strong validation culture (incremental execution, QA cycle, delivery gates) — top marks for actionability and workflow clarity. Its weaknesses are moderate redundancy across long sections and a complete absence of progressive disclosure: reference-grade material (raw XML appendix, specialty topics, recipes) is monolithically inlined rather than split into one-level-deep reference files.
Suggestions
Split the 'Raw-set XML appendix (L3 patterns)' and the 'Advanced / specialty topics' section (equations, comments/tracked changes, watermark) into a references/ file (e.g. references/advanced-ooxml.md), keeping SKILL.md an overview with clearly signaled one-level-deep pointers — this directly addresses the lowest-scoring dimension.
Move the 'Report-level recipes' (a)–(f) into a recipes reference file and keep one representative example inline, cutting SKILL.md length while preserving copy-paste access to the patterns.
Deduplicate repeated guidance to improve conciseness: consolidate the duplicated page-break rules ('Forcing page breaks' vs the pitfalls table row) and the updateFields instructions repeated across TOC, Fields, QA, and the XML appendix into one authoritative location cross-referenced from the others.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense, assumes Claude's competence (no filler explaining what Word/documents are; the ZIP-of-XML mental model is tool-specific knowledge), and nearly every command earns its place — but there is real redundancy across ~560 lines: page-break rules appear both in 'Forcing page breaks' and again in the pitfalls table, and updateFields guidance is repeated in TOC, Fields, QA, and the XML appendix. This matches 'efficient; minor instances of over-explanation that could be trimmed' rather than the lean 5 anchor. | 4 / 5 |
Actionability | Every section gives copy-paste-ready, complete `officecli` commands with exact props and values (e.g. `--prop pbdr.bottom="single;6;2E75B6"`, `--prop fieldType=mergefield --prop name=customer_name`), covering the common cases (headings, tables, lists, fields, TOC, footers) plus validated report-level recipes. This matches the fully-executable top anchor. | 5 / 5 |
Workflow Clarity | The Common Workflow is a clearly sequenced six-step process with explicit checkpoints ('After each structural op, `get` it back'), and QA is a bug-hunt cycle with a fix-then-rerun-full-cycle feedback loop plus a copy-paste Delivery Gate that REJECTs on any failure. For a file-mutating, batch-prone tool this is the top anchor: clear sequence, explicit validation, and error-recovery loops. | 5 / 5 |
Progressive Disclosure | Section structure and skip signals are good ('Advanced / specialty topics (skip if you are writing a report)'), but there are no bundle files at all: the L3 raw-set XML appendix, equations/comments/watermark specialty topics, and the report-level recipes are all inlined in a single ~560-line SKILL.md when they clearly belong in one-level-deep reference files. This sits between the 'minimal structure, content that belongs in separate files is inlined' (2) and the well-split 4/5 anchors — structure is solid but nothing is offloaded, so 3. | 3 / 5 |
Total | 17 / 20 Passed |