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.
A well-structured instruction skill: clear trigger scoping, an ordered 8-step workflow with reference-module integration, a mandatory output structure, and excellent progressive disclosure — all seven referenced bundle files exist and are explicitly mapped to workflow stages. The main cost is redundancy: prohibitions and goals are restated across four near-duplicate sections (Core Function, Hard Rules, What This Skill Should Not Do, Quality Standard), and the retrieval step lacks concrete tooling guidance.
Suggestions
Consolidate the overlapping directive sections — merge 'What This Skill Should Not Do' and 'Quality Standard' into 'Hard Rules' (or the Core Function lists), since rules like 'never treat technical complexity as methodological rigor' and 'prioritize consequential over common gaps' currently appear 3–4 times each.
Make Step 2 concrete: specify what retrieval tools or sources to use (e.g., which literature-search tool, how many anchor papers, how to weight the four source-type priorities) so the retrieval phase is executable rather than directional.
Add one brief worked example of a gap-map entry (label, class, where it appears, what it weakens, recurrence, resolved status) to make Section C's structure immediately copyable.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient — no filler explaining concepts Claude already knows, and the directive style is tight — but there is noticeable redundancy across sections: "This skill should not", "Hard Rules" (rules 3, 4, 7, 8), "What This Skill Should Not Do", and "Quality Standard" each restate the same prohibitions and goals three to four times (e.g., 'Never treat technical complexity as methodological rigor' appears in the description, Hard Rule 3, the not-do list, and effectively the Quality Standard). This matches 'mostly efficient but includes some unnecessary explanation or could be tightened' (3) rather than the 'minor instances' of score 4. | 3 / 5 |
Actionability | As an instruction-only skill it gives concrete, specific guidance: a validated input format with an explicit out-of-scope redirect template, an ordered 8-step execution procedure, classification categories for each step, a mandatory A–I output structure, and concrete trigger examples. It falls short of fully executable guidance (5) mainly because Step 2 ('Retrieve Method-Relevant Literature') never specifies what retrieval tooling or search strategy to use, and there is no worked example of a gap-map entry. This fits 'mostly executable guidance... with minor gaps' (4). | 4 / 5 |
Workflow Clarity | The 8 steps run in explicit order, each names its reference module, and Step 8 is a pre-finalization self-check with a concrete checklist. It misses score 5 because the self-critical review has no feedback loop (no 'if a check fails, return to Step N and revise' instruction), so checkpoints are present but the error-recovery loop is implicit. There are no destructive or batch operations, so no capping applies. | 4 / 5 |
Progressive Disclosure | The SKILL.md body is a well-organized overview and every one of the 7 files in references/ is real (verified on disk) and explicitly signaled with a one-level-deep pointer that states exactly when to use it (e.g., 'references/method-gap-taxonomy.md → use when classifying method gaps in Sections C–F'), plus section-level mapping to the output structure. This matches the score-5 anchor: clear overview, well-signaled one-level-deep references, easy navigation. | 5 / 5 |
Total | 16 / 20 Passed |