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 dense, accurate, and executable reference: concrete code for every feature, exact numeric limits, clear do/don't examples, and well-sequenced platform lifecycles with an error-to-fix table. Its main weaknesses are token efficiency at the margins (generic browser reconnect advice, billing detail, meta-comments) and the absence of any progressive disclosure — everything, including API reference tables, is inlined in one long file.
Suggestions
Move reference-grade material into bundle files (e.g. references/context-api.md for the Context fields and Types list, references/configuration.md for the config key table, references/monitoring.md for the logs walkthrough) and link to them from short sections in SKILL.md.
Cut the client-reconnect subsection (EventSource auto-reconnect is standard browser knowledge Claude already has) and trim the GA billing paragraph to the one actionable fact, keeping only the 1012 close-code convention.
Add a short numbered 'first deploy' sequence near the top (npm install @netlify/server → create netlify/server/index.ts → netlify dev → deploy) so the end-to-end workflow is explicit rather than implied across sections.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense with platform-specific facts Claude cannot know elsewhere ('Outbound connections are closed after 60 s without data... TCP keepalives don't count', 'listen within 10 s', 'Closed sockets with code 1012'), but a few passages restate generic knowledge ('Browser WebSocket doesn't reconnect on its own; EventSource does') and the GA billing paragraph plus trailing gap comments add tokens without runtime value. It sits above anchor 3 ('some unnecessary explanation') but short of anchor 5's 'every token earns its place'. | 4 / 5 |
Actionability | Nearly every section provides copy-paste-ready, complete code or commands with specific values: the handler export, Express/Fastify examples, 'idleTimeoutMillis: 30_000', upgradeWebSocket usage, lifecycle hooks, cache headers, config TOML, and 'netlify logs --source server --follow'. The error table maps each status (429/500/501/502) to a concrete fix, fully matching the anchor-5 'copy-paste ready... common cases' example. | 5 / 5 |
Workflow Clarity | Multi-step processes are clearly sequenced: startup steps 1-4 ('PORT is set' → import → onStart → inspect export), shutdown sequence 1-7, and the routing order 1-5, plus a 'Server gets no requests?' cause checklist and a status→fix error table that act as feedback loops. It falls short of anchor 5 because there is no single end-to-end 'first deploy' sequence tying install, entry file, local dev, and deploy together. | 4 / 5 |
Progressive Disclosure | No bundle files exist (no references/, scripts/, or assets/), so all ~360 lines live inline in SKILL.md, including reference-grade material — the Context API field list, the exported Types enumeration, the configuration key table, and the monitoring/logs walkthrough — that clearly belongs in separate files. Sections and headers are well organized (above anchor 2's 'minimal structure'), but the situation matches anchor 3: 'content that should be separate is inline', and the score-5 exception only applies to skills under 50 lines. | 3 / 5 |
Total | 16 / 20 Passed |