Content
78%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 an unusually tight, high-signal instruction set: lean to the point of telegraphic, with concrete sources, a checkable four-rule gate, and a fully specified output template. Its main defects are a broken external reference (author-gate.json is missing from the bundle, undermining the author gate and actionability), unspecified commands for two of the six sources, and an execution order that must be assembled from section layout.
Suggestions
Ship author-gate.json beside SKILL.md (with the establishedOwner and popularRepo entries rule 4 depends on) or inline its content, since the referenced file is absent from the bundle and the author gate cannot run as written.
Give concrete invocations for sources 3–4, e.g., the actual `gh api search/repositories?q=<need>` query and the npm/PyPI/crates.io lookup URLs, matching the specificity of the MCP and HN Algolia entries.
State the end-to-end sequence explicitly (skip check → search sources in order → gate each candidate on rules 1–4 → rank → output) and say what happens when a candidate fails (advance to the next candidate or source).
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The ~68-line body is relentlessly lean — every line is operative instruction ('Memory never fills in', 'Newness is no criterion: show age, never gate on it', the four-rule table, the output template), with zero explanation of concepts Claude already knows and no padding; it matches 'every token earns its place'. Not a 4 because no over-explained or trimmable passage was found. | 5 / 5 |
Actionability | Mostly executable: concrete commands and URLs are given ('npx skills find <need>', the MCP registry URL, the HN Algolia URL, the read-only gh/curl constraints), the rules are checkable criteria, and the output block is a copy-paste template. It falls short of fully-executable because sources 3–4 ('GitHub through gh api', 'The npm, PyPI and crates.io registries') name no specific query command or URL, and the `author-gate.json` data the author check depends on is not in the bundle. | 4 / 5 |
Workflow Clarity | A clear sequence exists in execution order (write memory pick → Skip check → Sources in order → four-rule gate with author check → Rank → Output) with explicit checkpoints ('Check all four before rejecting a candidate, and list every rule that fails', 'Check each name against its registry before naming it') and early exits ('say which and stop', 'say so and stop'). Not a 5 because the end-to-end order must be inferred from section layout rather than stated, and there is no instruction for what to do after a candidate fails the rules (move to the next candidate or source). | 4 / 5 |
Progressive Disclosure | Sections are well organized for a single ~68-line file, but the body's only external reference — 'establishedOwner in author-gate.json, beside this file' — points to a file that does not exist in the bundle (no references/, scripts/, assets/, or author-gate.json anywhere), so the one reference it makes is broken and rule 4's (a)/(b) checks cannot be executed as written. Scored against the actual bundle structure per the guideline, this lands at 'some structure but could be better organized' rather than the 'minor organization gaps' of a 4, since the missing data file is more than a minor gap. | 3 / 5 |
Total | 16 / 20 Passed |