Conciseness

Mostly efficient — commands and the proxy flag table are tight — but several parts are trimmable: the intro paragraph explaining what the app is for, the '## Popular actions' heading whose body just repeats the `action list` command already shown in 'Searching for actions', and the informal aside on agent types ('Those will be used to adjust tooling to be used best with your harness'). Falls between anchor 3 ('some unnecessary explanation or could be tightened') and anchor 4 ('minor instances of over-explanation').

Actionability

Nearly every step has a copy-paste command ('npm install -g @membranehq/cli@latest', 'membrane connection ensure "https://ecartapi.com/" --json', 'membrane action run <actionId> --connectionId=CONNECTION_ID --input \'{"key": "value"}\' --json') plus a flag table for proxy requests, matching 'Mostly executable guidance; concrete code or commands with minor gaps'. Not 5 because there is no worked example with a real action id or intent query, so the placeholders CONNECTION_ID and QUERY are never exercised end-to-end.

Workflow Clarity

The sequence (install → login → connection ensure → poll state → search actions → run → proxy fallback) is present with state-based checkpoints ('READY', 'CLIENT_ACTION_REQUIRED', 'CONFIGURATION_ERROR or SETUP_FAILED — check the error field', 'Keep polling until state is no longer BUILDING'). But the doc references 'skip to Step 2' and a '#### 1b' heading while no section is ever labeled Step 1, 2, or 1a — the step numbering is incoherent — and it switches between 'membrane' and 'npx @membranehq/cli@latest' invocations without reconciliation. Falls between anchors 3 and 4.

Progressive Disclosure

No bundle files exist (no references/, scripts/, or assets/ directories), so all content is inline in a ~140-line body organized under clear headings (Install, Authentication, Connecting, Searching for actions, Running actions, Proxy requests, Best practices) with an external link to official docs — matching 'Good structure; most content is appropriately placed; minor organization gaps'. Not 5 because the empty '## Popular actions' section that nests '### Running actions' is an organization gap, and the connection-state handling detail is borderline content that could live in a reference file.

Specificity

Names the domain ('API to every Ecommerce Cart and Marketplace integration') but the actions are pure boilerplate — 'Manage data, records, and automate workflows' — with no concrete verbs such as list orders, sync products, or fetch customers. This matches the anchor 'Names the domain but actions are minimal or generic'; not 3 because no concrete actions are actually listed.

Completeness

Both parts are explicit: the what ('integration... Manage data, records, and automate workflows') and the when ('Use when the user wants to interact with API to every Ecommerce Cart and Marketplace data'). The 'when' is tautological — it restates the skill name rather than giving concrete trigger phrases — matching 'Has both what and when; when could be more explicit or specific'. Not 3 because the 'Use when' clause is explicit rather than weakly implied; not 5 because neither part names concrete trigger phrases.

Trigger Term Quality

'Ecommerce', 'Cart', and 'Marketplace' are relevant natural keywords, but the common variations users would actually say — orders, products, customers, listings, or platform names like Shopify/Amazon/eBay — are absent, matching 'Some relevant keywords but missing common variations or synonyms'. Not 4 because keyword coverage stops at restating the domain name.

Distinctiveness Conflict Risk

The named niche makes wrong-skill triggering unlikely — 'Use when the user wants to interact with API to every Ecommerce Cart and Marketplace data' scopes it tightly to that API. Not 5 because the generic 'Manage data, records, and automate workflows' phrasing is shared boilerplate that overlaps with sibling Membrane connector skills for other apps, relying solely on the product name for distinction.