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.
This is a strong, highly actionable integration skill: nearly every instruction is an executable command, the connection lifecycle includes state-based checkpoints and re-polling feedback loops, and the action/proxy tables make the API surface immediately usable. The main weaknesses are minor: a few sentences of skippable platform boilerplate, broken 'Step 2' cross-references, and no progressive disclosure — all content lives in one file with no reference splitting, even though the Membrane setup sections are shared boilerplate a bundle could hold.
Suggestions
Fix the incoherent step numbering: either label 'Searching for actions' as 'Step 2' (and add a matching 'Step 1') or drop the 'skip to Step 2' phrasing entirely, since readers are sent to a section that doesn't exist by that name.
Move the connection-state reference (READY / CLIENT_ACTION_REQUIRED / CONFIGURATION_ERROR handling) and the proxy flag table into a references/ file (e.g. connections.md, proxy.md) and point to them from SKILL.md, keeping the body as a lean overview.
Trim the marketing-flavored opener ('E-commerce businesses of all sizes use AfterShip to improve the post-purchase experience') and the empty 'AfterShip Overview' bullet list down to one factual sentence about what AfterShip exposes (trackings, couriers).
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly lean: install/auth/connect commands, a popular-actions table, and a flag table with little prose around them. Minor over-explanation remains — the opening paragraph ('E-commerce businesses of all sizes use AfterShip to improve the post-purchase experience') and the near-empty 'AfterShip Overview' section list Trackings/Couriers without saying anything actionable. Not a 3 because padding is limited to these few sentences; not a 5 because those sections are clearly trimmable. | 4 / 5 |
Actionability | Every section is copy-paste ready: `npm install -g @membranehq/cli@latest`, `membrane connection ensure "https://www.aftership.com/" --json`, `membrane action run <actionId> --connectionId=... --input '{...}'`, plus concrete flag tables for the proxy and a named action table with keys and descriptions. Common cases (login, connect, search, run, proxy) are each covered by an executable example. | 5 / 5 |
Workflow Clarity | There is a clear sequence (install → login → connection ensure → poll state → search actions → run/proxy) with real checkpoints: poll `--wait` until state leaves `BUILDING`, branch on `READY`/`CLIENT_ACTION_REQUIRED`/`CONFIGURATION_ERROR`, and re-poll after the user authenticates — an explicit feedback loop. Minor gap: the text says 'skip to Step 2' twice but no section is labeled Step 2 (and '1b' has no matching '1a'), leaving the numbering incoherent. Not a 5 because of that broken step numbering and the absence of any verification step after running a mutating action. | 4 / 5 |
Progressive Disclosure | No bundle files exist (references/, scripts/, assets/ are all absent), so everything — including generic Membrane setup (install, login, connection lifecycle states) that applies to any Membrane-based skill — is inlined in a single ~145-line file. Structure within the file is good (clear headers, tables), but there is no one-level-deep reference to split the shared CLI boilerplate from the AfterShip-specific content, matching anchor 3 ('some structure but could be better organized; content that should be separate is inline'). Not a 4 because the proxy flag table and connection-state reference are reference material sitting inline; not a 2 because the file is well-sectioned and easily navigable. | 3 / 5 |
Total | 16 / 20 Passed |