Content
11%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 a hollow shell: generic boilerplate sections, an abstract instruction list, and a phase heading with no steps, pointing exclusively at 16 sub-skill files and a playbook that do not exist in the bundle. A reader following any reference finds nothing, and nothing in the body itself is actionable. The only structural virtue is that it is short and points one level deep.
Suggestions
Add real executable substance to the Instructions: concrete first steps, specific commands or code patterns for instrumentation/experiment setup, and explicit validation checkpoints (e.g. verify analytics events fire before rollout, check experiment power calculations before launch).
Fix the workflow: either describe the full phase sequence (Phase 1 through N) with ordered steps and validation gates, or drop the phase framing entirely — currently 'Phase 1' contains no steps and implies missing phases.
Repair progressive disclosure: ship the referenced ./sub-skills/*.md and resources/implementation-playbook.md files (or remove the dead links), and give each link a one-line description of what it covers so navigation is possible.
Cut the filler: delete the bracketed 'Extended thinking' meta-paragraph and the tautological use/don't-use sections, which consume tokens without adding guidance.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~100-word bracketed "[Extended thinking: This workflow orchestrates a comprehensive data-driven development process...]" meta-paragraph, the tautological "Use this skill when / Do not use this skill when" sections ("The task is unrelated to data-driven feature development"), and filler bullets like "Apply relevant best practices and validate outcomes" are padded sections adding no information — matching 'Noticeably verbose; several unnecessary explanations or padded sections'. Not 1 because it never explains concepts Claude already knows; not 3 because the majority of the body is filler rather than an occasional stray sentence. | 2 / 5 |
Actionability | The Instructions are entirely abstract — "Clarify goals, constraints, and required inputs", "Apply relevant best practices and validate outcomes", "Provide actionable steps and verification" describe rather than instruct, with no concrete steps, commands, or examples — matching 'Entirely vague or abstract; no concrete code or commands'. The single concrete pointer ("open resources/implementation-playbook.md") targets a file that does not exist in the bundle, so it provides no executable guidance. | 1 / 5 |
Workflow Clarity | "## Phase 1: Data Analysis and Hypothesis Formation" is a header with no steps under it — the phase contains only an unrelated list of 16 sub-skill links, no Phase 2+ exists despite the numbering implying more, and the sole validation mention ("Apply relevant best practices and validate outcomes") is generic. This matches 'Steps missing or incoherent; no sequence; no validation for risky operations'. | 1 / 5 |
Progressive Disclosure | Scored against the actual bundle: the bundle contains only SKILL.md — neither ./sub-skills/ (16 linked files) nor resources/implementation-playbook.md exist, so every reference is broken. The textual structure is one level deep and the body is a thin pointer file (the right shape), but the links go nowhere, are bare numbered titles with no description of content, and sit under a mislabeled phase header — matching 'Minimal structure; ... or references are buried'. Not 1 because the intended navigation is flat and legible rather than deeply nested; not 3 because broken paths mean the references cannot actually be followed. | 2 / 5 |
Total | 6 / 20 Passed |