Content
68%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 lean, project-specific skill with a concrete executable auth example and specific failure-mode guidance, notably the scan-timing workflow around error code 70. Its main weaknesses are the absence of a sequenced integration workflow and directive-only guidance (backoff, caching, scan polling) for the patterns most central to netrunner.
Suggestions
Present the core netrunner integration as a numbered workflow with explicit checkpoints, e.g. 1) configure .env, 2) call startScan after downloads, 3) poll getScanStatus, 4) only then query metadata, 5) handle error.code 70 by re-polling rather than failing.
Add small executable snippets for the scan loop (startScan + getScanStatus polling) and exponential backoff, since these are the skill's most distinctive guidance but currently have no code.
Trim tangential content: drop the Python/Go library entries and the per-agent Invocation section (or move them to a reference file) to tighten the token budget.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is lean and does not explain concepts Claude already knows (no 'what is Subsonic' padding); every section carries project-specific information. Minor trimmable padding exists — the Python/Go library entries and the per-agent 'Invocation' section are tangential to the TypeScript-focused netrunner use case — keeping it below the 'every token earns its place' anchor of 5 but above 3, which requires unnecessary explanations. | 4 / 5 |
Actionability | The TypeScript auth snippet is concrete and executable (env-var config, createSubsonicAPI, getArtists) and failure modes cite exact error codes ('error.code === 70'), matching 'mostly executable guidance with minor gaps'. Not a 5 because the most netrunner-specific guidance — startScan/getScanStatus polling, exponential backoff, metadata caching — is directive-only with no executable examples. | 4 / 5 |
Workflow Clarity | The scan workflow embeds a real checkpoint ('Use startScan after new downloads; poll getScanStatus before querying new files'), but the overall integration flow is scattered across Auth Pattern, Key Patterns, and Failure Modes rather than presented as a sequenced workflow — the sequence is present but implicit, matching the 3 anchor. Not a 4 because there is no clear end-to-end sequence for the primary task (set up auth, query metadata, handle scan timing). | 3 / 5 |
Progressive Disclosure | Sections are well-organized (Purpose, Libraries, Auth, Patterns, Inputs/Outputs, Safety, Failure Modes, References) and the References section is clearly signaled with one-level-deep external links; the body stands alone at ~80 lines. Not a 5 because the skill exceeds the under-50-line simple-skill bar and endpoint/error-code details would eventually belong in a reference file; not a 3 because no content that clearly belongs in a separate file is inlined. | 4 / 5 |
Total | 15 / 20 Passed |