Content
81%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, safety-conscious skill body with strong sequencing, explicit validation checkpoints, and concrete commands. Main weaknesses are minor: some repeated safety messaging and an inline QMD section that could be split into a reference file.
Suggestions
Provide literal copy commands (e.g. cp -r / rsync) for the archive and restore copy steps instead of describing them as actions, to make Mode 1/2/3 fully copy-paste executable.
Factor the 'QMD Refresh After Live Wiki Changes' section into a separate reference file and keep a short pointer in SKILL.md, reducing inline length and avoiding repetition of the QMD status strings.
Dedupe the safety messaging so each rule appears once in 'Safety Rules' rather than being restated in each mode's steps, tightening token use without losing the guardrail.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly lean and purposeful for a destructive multi-mode skill, but safety messaging is repeated across sections and the QMD report enumeration (six status strings) plus the archive tree diagram could be trimmed. Not a 5 because not every token earns its place; not a 3 because the prose is efficient and avoids explaining concepts Claude already knows. | 4 / 5 |
Actionability | Provides concrete QMD commands ('${QMD_CLI:-qmd} update/embed/ls/get'), exact directory paths, and log-entry templates, but the archive-copy steps are described as actions ('Copy all category directories...') rather than literal commands, leaving a minor gap. Above the pseudocode anchor, below fully copy-paste-ready. | 4 / 5 |
Workflow Clarity | Each of three modes has numbered steps with explicit validation checkpoints (confirm intent, archive-before-destroy, warn-before-overwrite), a Safety Rules checklist, and error-recovery guidance (restore from archive; do not roll back on QMD failure). The destructive-operation cap does not apply because validation is present, fitting the top anchor. | 5 / 5 |
Progressive Disclosure | Well-organized with clear section headers and one-level-deep external skill references (llm-wiki/SKILL.md, ingest skills), no nested refs. Held at 4 rather than 5 because the lengthy inline QMD Refresh block is somewhat ancillary and could be externalized into a reference file. | 4 / 5 |
Total | 17 / 20 Passed |