Content
60%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 well-structured with a genuine safety-first workflow (preview, -WhatIf, non-destructive defaults, hash verification) and mostly executable PowerShell examples. Its weaknesses are a broken/dead code snippet in the first example and Python split/merge commands whose flags do not match the bundled scripts (the text even warns they may be wrong), plus redundancy between Key Features, Implementation Details, and When to Use.
Suggestions
Fix the split/merge commands to match the bundled scripts' actual CLIs (positional input path, --part-size-mb as integer MB, --output-dir/--prefix/--pad-width for split; positional parts-or-directory and --output/--overwrite for merge) and delete the "CLI flags may vary" hedge.
Remove the dead initial filter block in Example 1 and fix the exclude-dir filter so it escapes the actual directory names ($_) rather than the literal string "\$_\".
Merge the duplicated "Key Features" and "Implementation Details" bullets (conflict handling, verification, rule schema) into one section, and move the long PowerShell organize example or the rule-format details into references/rule-schema.md to slim SKILL.md.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient but has clear slack: the first PowerShell example contains a dead, broken filter block ("($Include | ForEach-Object { $_ }) -contains $_.Extension -eq $false") that is immediately superseded by a second filter; "Key Features" and "Implementation Details" substantially repeat each other (conflict handling, verification, rule schema); and "When to Use" restates the description. This matches the 3 anchor (mostly efficient, some unnecessary explanation that could be tightened) — not 2, since there is no concept-explaining padding, and not 4, since the dead code and duplicated sections are more than minor. | 3 / 5 |
Actionability | There is substantial concrete guidance (executable PowerShell for plan/preview/conflict-rename, a hash-manifest backup, and compress commands), but key details are wrong or missing: the split/merge commands do not match the shipped scripts ("--input"/"--part-size 100MB"/"--input-dir" vs. the actual positional input, "--part-size-mb" int, and positional inputs), and the body even hedges "CLI flags may vary... adapt the commands", plus the exclude filter escapes a literal "\$_\" rather than the variable names. This fits the 3 anchor (concrete guidance but incomplete, missing key details) — not 4, because the documented Python commands are not runnable as written against the bundle's own scripts. | 3 / 5 |
Workflow Clarity | The batch/destructive-operation cap is not triggered: the content mandates a preview manifest before any write, uses -WhatIf dry-run, defaults to non-destructive rename/skip with overwrite only after explicit confirmation, and specifies verification (counts/sizes plus SHA256 manifests for critical backups). That satisfies the 4 anchor (clear sequence with most checkpoints present, minor validation gaps) — e.g., a post-copy count/size comparison command is described but not shown, and there is no validate→fix→retry feedback loop like the 5 anchor's "If errors: fix and re-validate". | 4 / 5 |
Progressive Disclosure | Measured against the actual bundle: references/rule-schema.md, scripts/split_file.py, and scripts/merge_parts.py all exist and are clearly signaled one level deep ("see references/rule-schema.md", "Use references/rule-schema.md as the canonical format", Dependencies listing the scripts). Structure is good (When to Use, Key Features, Dependencies, Example Usage, Implementation Details). It falls short of the 5 anchor only because the ~60-line inline PowerShell example and the rule-planning details in Implementation Details could partly live in a reference file, leaving SKILL.md a leaner overview. | 4 / 5 |
Total | 14 / 20 Passed |