Content
72%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 highly actionable, command-rich integration skill body with strong feedback loops in the connection workflow. The main improvements are tightening a few explanatory passages, fixing step numbering, and splitting dense reference material into a bundled file.
Suggestions
Tighten the prose around auth and connection setup to remove phrases that restate what the commands already show.
Normalize the step numbering (label 'Step 1' explicitly and replace the '1b' / unlabeled 'Step 2' references) so the workflow reads as one coherent sequence.
Move the clientAction enum and proxy flag table into a references/REFERENCE.md and link from the body to reduce inline bulk.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient — the body is dominated by direct CLI commands — but includes a few explanatory passages ('focus on the integration logic rather than auth plumbing', the full clientAction enum, 'fastest way to get a connection') that could be trimmed. | 4 / 5 |
Actionability | Copy-paste-ready executable commands throughout (npm install, membrane login, connection ensure, action run, request) with parameter flags and a flag reference table covering the common integration cases. | 5 / 5 |
Workflow Clarity | Clear install → login → connect → wait → search → run sequence with genuine feedback loops (poll --wait, READY/BUILDING/CLIENT_ACTION_REQUIRED/CONFIGURATION_ERROR states, re-poll after user action); minor gaps from inconsistent step numbering ('1b', 'Step 2' without a labeled Step 1). | 4 / 5 |
Progressive Disclosure | Single SKILL.md with well-organized sections and no nested references, but with no bundle files the inline clientAction enum and proxy flag table are bulk that a separate REFERENCE.md could absorb. | 4 / 5 |
Total | 17 / 20 Passed |