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.
The body is highly actionable with concrete Membrane commands and useful tables, but loses points for explaining Stripe concepts Claude already knows and for offering no verification guidance around destructive operations. Structure is good but everything is inlined into one long file.
Suggestions
Remove or drastically shorten the opening paragraph explaining what Stripe is — Claude already knows this; lead instead with the actionable Membrane workflow.
Add an explicit verification/confirm-before-acting step for destructive actions (delete-product, update-product/subscription/price), e.g. read-then-write or a dry-run check, and resolve the dangling "skip to Step 2" reference.
Move the popular-actions and proxy-flags tables into a separate reference file (e.g. ACTIONS.md) referenced from the body to reduce inline length and improve progressive disclosure.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The opening paragraph ("Stripe is a payment processing platform that enables businesses to accept online payments...") explains what Stripe is — a concept Claude already knows and which the rubric penalizes — alongside a few fluffy sentences ("so you can focus on the integration logic rather than auth plumbing"); the rest is mostly efficient commands and tables. | 3 / 5 |
Actionability | The body provides fully executable, copy-paste-ready commands ("membrane connection ensure", "membrane action list --intent", "membrane action run", "membrane request") plus a popular-actions table with keys and a proxy-flags table, covering the common cases as anchor 5 requires. | 5 / 5 |
Workflow Clarity | The connection sequence has explicit state checkpoints (READY/BUILDING/CLIENT_ACTION_REQUIRED) and a poll-after-action feedback loop, but the skill enables destructive actions (delete-product, update-product/subscription/price) with no verification guidance, and the "skip to Step 2" reference is dangling — the rubric's destructive-operation cap prevents scoring above 3. | 3 / 5 |
Progressive Disclosure | Content is well-organized into clear sections, but with no bundle files everything lives in a single ~180-line file (exceeding the 50-line simple-skill exception) and reference-like material (the popular-actions and proxy-flags tables) is inlined rather than split into a separate reference file. | 4 / 5 |
Total | 15 / 20 Passed |