Content
88%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 strong operational skill body: executable, placeholder-resolved commands with complete options and usage examples, a sequenced workflow with dry-run validation and explicit error-recovery branches, and correct separation of implementation into the bundled script. The only deductions are trimmable meta-guidance about user-input tooling and inlined source-map internals that belong in the script or a reference file.
Suggestions
Compress the "User Input Tools" section to a single line (e.g., 'Ask the user via the runtime's user-input tool if available; otherwise numbered plain-text questions, batched when possible') — Claude already knows how to prompt users, and the current ~9 lines are the largest source of avoidable tokens.
Move the "Source-Map Path Restoration" internals (sourceRoot resolution, ../-segment policy, URL stripping, __unknown fallback) into comments in scripts/main.ts or a reference file; SKILL.md only needs the outcome — restored/ is rebuilt from embedded sourcesContent and skipped maps are listed in extract-report.json warnings.
Trim the Usage block's overlapping examples with Workflow step 2 (the plain app-name invocation appears in both) to remove duplication and reduce token cost.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and operational (options table, output layout, platform paths) with no explanation of concepts Claude already knows, but the "User Input Tools" section spends ~9 lines on meta-guidance about prompting mechanics Claude can handle, and the Usage block duplicates the Workflow's command. This fits 'efficient; minor instances of over-explanation that could be trimmed' rather than the every-token-earns-its-place anchor at 5. | 4 / 5 |
Actionability | Commands are fully executable once the two explicitly-defined placeholders are substituted ("Replace {baseDir} and ${BUN_X} with actual values", with the bun/npx resolution rule given), and the Usage block covers the common cases: app name, macOS .app path, Windows install dir, --asar override, --output, and --json. Copy-paste ready with specific examples; no gaps that would drop it to 4. | 5 / 5 |
Workflow Clarity | The 4-step workflow has explicit validation and feedback loops: "run with --dry-run first" as a pre-check, and step 3's error-recovery branches ("Multiple matches → ... re-run with the chosen absolute path", "Existing non-empty output dir → ... --force or pick a new --output"), plus a result report with counts and warnings. Although the operation is batch-like, validation is present (dry-run, safety refusals, extract-report.json warnings), so the missing-validation cap at 3 does not apply. | 5 / 5 |
Progressive Disclosure | The bundle structure is well used: all implementation lives in the clearly-signaled one-level-deep `scripts/main.ts` (verified to exist), and SKILL.md stays an operational overview. However, the "Source-Map Path Restoration" section inlines ~10 lines of script internals (path-resolution edge cases) that Claude does not need to run the skill, which fits 'good structure; minor organization gaps' rather than the fully clean split at 5. The unreferenced scripts/main.test.ts in the bundle is a minor navigational gap. | 4 / 5 |
Total | 18 / 20 Passed |