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.
A high-quality, information-dense body: nearly everything is platform-specific, executable, and organized around real failure modes with concrete fixes. The main structural weakness is that it is a single monolithic ~340-line file with zero progressive disclosure — the API reference, config tables, and monitoring details could live in one-level-deep reference files — and there are small patches of general-knowledge content that could be trimmed.
Suggestions
Split detailed reference material into one-level-deep bundle files (e.g., references/config.md for the configuration key table, references/context-api.md for the context fields and types, references/monitoring.md for the Server page and log-drain walkthrough), keeping SKILL.md as a lean overview with clearly signaled links.
Trim or shorten content Claude already knows, such as the client-reconnect example ("Browser WebSocket doesn't reconnect on its own; EventSource does") — keep only the platform-specific detail like closing with code 1012.
Add a brief end-to-end sequence (install @netlify/server → write entry file → export correctly → set config → verify locally with netlify dev → deploy → check the Server page/logs) so the reference sections hang off an explicit build-and-verify workflow.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dense platform-specific fact (entry-file detection table, 60 s outbound idle rule, shutdown sequence, error table, house rules) that Claude cannot know, with almost no padding or explanation of known concepts. Minor trimmable spots remain: the client-reconnect section largely restates general browser knowledge ("Browser WebSocket doesn't reconnect on its own; EventSource does") and the monitoring UI walkthrough narrates page layouts in detail. This matches "efficient; minor instances of over-explanation that could be trimmed" rather than 5's "every token earns its place". | 4 / 5 |
Actionability | Guidance is copy-paste ready throughout: working handler and framework starter code, executable waitUntil/WebSocket/SSE/hook examples with the exact API calls, a concrete config TOML, CLI commands with flags ("netlify logs --source server --since 30m"), and an error table mapping each status code to a specific fix. Common cases (Express, Fastify, Hono, ws, self-started servers) are each covered by example. This matches the fully-executable anchor; not 4 because there are no significant gaps in code completeness. | 5 / 5 |
Workflow Clarity | Key processes are explicitly sequenced with numbered lists (startup: PORT → import → onStart → export inspection; shutdown: 7 ordered steps with timeouts) and troubleshooting is checklist-driven ("Server gets no requests? Check these causes...", the Errors table, the 60 s connection-notice symptom→fix). However, there is no single end-to-end build→test→deploy→verify path; guidance is organized as reference sections rather than a sequenced procedure, and verification steps (e.g., confirming a deploy started) are implicit. This fits "clear sequence with most checkpoints present; minor validation gaps"; the destructive/batch cap does not apply since the skill involves no batch/destructive operations. | 4 / 5 |
Progressive Disclosure | The file is well-organized with clear section headers, but there are no bundle files at all — references/, scripts/, and assets/ do not exist — so nothing is split out. Detailed reference material that plausibly belongs in separate files is inlined in the ~340-line body: the Context API field/type listing, the full configuration key table, the monitoring/logs UI walkthrough, and log-drain details. With no external references and substantial inlined reference content, this matches "some structure... content that should be separate is inline" rather than 4's "most content is appropriately placed". | 3 / 5 |
Total | 16 / 20 Passed |