Content
71%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 highly actionable — every command is executable with exact flags, JSON shapes, and failure diagnostics — and the preflight/search/maintenance flow is well checkpointed. Its weaknesses are structural and editorial: ~290 lines of inlined reference-grade detail that belongs in bundled reference files, a dangling docs/minsync-setup.md pointer, and several over-long passages that could be tightened without losing information.
Suggestions
Move the calculator operations, filee/dupey conversion details, and automations.json specification into files under references/ (e.g. references/calculator.md, references/conversions.md, references/automations.md), leaving SKILL.md as a lean overview with clearly signaled one-level-deep links.
Fix or remove the dangling "docs/minsync-setup.md#embedding-backends" reference — no docs/ directory exists in the bundle — by shipping that file or pointing to an existing location.
Tighten the multi-clause passages (the search `locations` bullet, the timezone-abbreviation rules, and the evidence-field list) into shorter bullets or a table so each token earns its place.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense and almost entirely skill-specific operational detail Claude could not know (config keys, diagnostics names, tool semantics), so it avoids the 'explains known concepts' failure, but sections like the calculator operations list, the timezone-abbreviation rules, and the multi-clause `locations` bullet run long and could be tightened, matching 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the minor-trim profile of anchor 4. | 3 / 5 |
Actionability | Commands are copy-paste ready throughout ("autorag search \"...\" --top-k 5 --json --debug", "autorag evidence <sessionId> --result 1 --json", the full maintenance command block) and concrete JSON examples cover common cases like the automation rule, matching 'fully executable; copy-paste ready code or commands; specific examples cover the common cases'; anchor 4 would imply missing key details, but flags, fallbacks, and config keys are all specified. | 5 / 5 |
Workflow Clarity | The Preflight → Search → Maintenance flow is clearly sequenced with explicit checkpoints (preflight config check plus `status`/`health --json`, `--debug` diagnostics for citation and reference verification, automation loop guards and run-rate disabling, and a recovery path 'use autorag-setup rather than guessing'), fitting 'clear sequence with most checkpoints present'; it falls short of anchor 5 because the destructive/batch paths (file conversions, automation agent writes) lack a systematic validate-then-fix feedback loop rather than scattered guards. | 4 / 5 |
Progressive Disclosure | Headers organize the ~290-line body, but no bundle files exist (no references/, scripts/, or assets/), so deep reference material — calculator semantics, filee/dupey conversion details, the automations.json spec — is inlined in SKILL.md where it belongs in separate files, and the single outward reference ("docs/minsync-setup.md#embedding-backends") is buried in the Rules section and dangling (no docs/ directory in the bundle), matching 'some structure but could be better organized; references present but not clearly signaled; content that should be separate is inline'. | 3 / 5 |
Total | 15 / 20 Passed |