docx

Comprehensive document creation, editing, and analysis with support for tracked changes, comments, formatting preservation, and text extraction. When Claude needs to work with professional documents (.docx files) for: (1) Creating new documents, (2) Modifying or editing content, (3) Working with tracked changes, (4) Adding comments, or any other document tasks

1.14x

Quality

81%

Does it follow best practices?

Impact

95%

1.14x

Average score across 2 eval scenarios

Securityby

Medium

Suggest reviewing before use

Quality

Content

75%Weight 40%Scale 1-5

Reviews 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 skill file: the decision tree, numbered workflows, executable commands, and explicit final verification make it easy to follow. The main weaknesses are redundant repetition of the batching strategy across the redlining section, a missing error-recovery loop after verification, and script paths that don't match the actual bundle layout.

Suggestions

Consolidate the batching guidance into one place: the "Batching Strategy" paragraph, step 2's "Batch organization" list, and step 4's "Suggested batch groupings" repeat the same 3-10-change rule and near-identical grouping schemes (by section/type/proximity) three times — state it once and reference it from the steps.

Add an explicit error-recovery loop to the final verification step, e.g. 'If grep still finds an original phrase or an unintended change appears, return to the failing batch, fix the script, re-run, and re-verify' — this is especially important for a batch/destructive workflow.

Align script paths with the actual bundle layout (e.g. `scripts/unpack.py` / `scripts/pack.py` rather than `ooxml/scripts/...`), and mention the shipped Document library files (scripts/document.py, scripts/templates/) directly so the body's navigation matches the bundle.

Dimension	Reasoning	Score
Conciseness	The body is mostly efficient with concrete commands and no concept over-explanation, but the batching guidance is repeated three times: "Group related changes into batches of 3-10 changes" in the intro, "Batch organization (group 3-10 related changes per batch)" in step 2, and again "batch size of 3-10 changes works well" plus a near-verbatim repeat of the same four grouping schemes in step 4. This redundancy is more than the 'minor instances' of level 4 but far from the pervasive padding of level 2.	3 / 5
Actionability	Fully executable, copy-paste-ready commands throughout: `pandoc --track-changes=all path-to-file.docx -o output.md`, `python ooxml/scripts/unpack.py <office_file> <output_directory>`, `soffice --headless --convert-to pdf document.docx`, `pdftoppm -jpeg -r 150 -f 2 -l 5 document.pdf page`, grep verification commands, a concrete GOOD/BAD tracked-changes XML example, and explicit dependency install commands. Specific examples cover the common cases.	5 / 5
Workflow Clarity	A decision tree routes tasks to clearly numbered workflows, and the redlining workflow includes per-batch testing and an explicit final verification step (pandoc conversion, grep for original/replacement phrases, unintended-change check). It falls short of level 5 because no error-recovery loop is specified when verification fails — no 'fix the batch and re-verify' guidance — and validation appears only at the end. Validation steps are present, so the destructive/batch cap of 3 does not apply.	4 / 5
Progressive Disclosure	The SKILL.md is a clear overview with well-signaled, one-level-deep references: "Read [`ooxml.md`](ooxml.md) (~600 lines)" and "Read [`docx-js.md`](docx-js.md) (~500 lines)" with explicit MANDATORY directives. Minor gaps remain relative to the actual bundle: referenced script paths ("ooxml/scripts/unpack.py") do not match the on-disk layout ("scripts/" containing document.py, utilities.py, templates/), and the bundle's actual files are not directly named or navigated from the body.	4 / 5
	Total	16 / 20 Passed

Description

88%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description that explicitly states both capabilities and trigger conditions with concrete, enumerated task types anchored on the .docx format. Its only weaknesses are minor: missing "Word" as a synonym and a slightly broad catch-all tail that leaves small overlap risk with sibling document skills.

Dimension	Reasoning	Score
Specificity	The description lists multiple concrete actions — "document creation, editing, and analysis with support for tracked changes, comments, formatting preservation, and text extraction" — and enumerates four specific task types (creating, modifying, tracked changes, comments), comprehensively covering the skill's workflows. It exceeds level 4's 'minor gaps in coverage' since the enumeration mirrors the full body scope.	5 / 5
Completeness	It explicitly answers both: what ("Comprehensive document creation, editing, and analysis with support for tracked changes, comments, formatting preservation, and text extraction") and when ("When Claude needs to work with professional documents (.docx files) for: (1) Creating new documents, (2) Modifying or editing content, (3) Working with tracked changes, (4) Adding comments") with concrete trigger phrases. Both are explicit, matching the level 5 anchor.	5 / 5
Trigger Term Quality	Good natural keyword coverage: ".docx files", "tracked changes", "comments", "creating new documents", "text extraction". It falls short of level 5 because common synonyms like "Word" or "Word documents" are missing, though coverage is well above the 'missing common variations' of level 3.	4 / 5
Distinctiveness Conflict Risk	It anchors clearly on ".docx files" with a distinct, enumerated task list, giving it a clear niche with only minor overlap risk against sibling office-document skills. It is not level 5 because the catch-all "or any other document tasks" broadens the trigger surface slightly beyond the .docx niche.	4 / 5
	Total	18 / 20 Passed

Validation

87%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 14 / 16 Passed

Validation for skill structure

Criteria	Description	Result
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning
relative_links	Relative link issues: 3 missing	Warning

	Total	14 / 16 Passed

Repository: netease-youdao/LobsterAI
Path: SKILLs/docx/SKILL.md
Commit: 791a352

Reviewed: 2 days ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.