Content
57%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 strong on executable commands — the auth, connection lifecycle, and proxy sections are concrete and usable — but it is undermined by broken navigation ('skip to Step 2' with no Step 2, '1b' without '1a'), an empty 'Popular actions' section, and an unexplained entity list that reads like template residue. Adding run-result verification and a few worked action examples would lift it substantially.
Suggestions
Fix the navigation: label the actual steps (or replace 'skip to Step 2' and the orphan '1b' heading with named sections), and remove or explain the bare Customer/Subscription/Invoice entity list.
Make 'Popular actions' earn its heading: list a few concrete example invocations with realistic action ids and --input payloads, including how to inspect the output field and handle errors after 'action run'.
Pick one CLI invocation style (membrane vs npx @membranehq/cli) and use it consistently; the current mix of lines 78 and 112 with the rest invites confusion.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient — commands come with tight, useful explanations (headless auth flow, connection-state decision tree, proxy flag table) — but there is padding: the bare entity bullet list (Customer/Subscription/Invoice/...) explains nothing, 'Popular actions' repeats the discovery command already shown in 'Searching for actions', and the agent-types aside is vague. Matches anchor 3; not 4, because several sections could be trimmed or removed outright; not 2, because it never lectures on concepts Claude already knows. | 3 / 5 |
Actionability | Nearly everything is copy-paste-ready bash: install, login (including the headless completion flow), 'connection ensure', polling with --wait, action search/run with --input JSON, and proxy requests with a flag table. Matches anchor 4 ('Mostly executable guidance... minor gaps'); not 5, because no worked example with a real action id and input is shown and the 'Popular actions' section lists none; not 3, because nothing shown is pseudocode or abstract. | 4 / 5 |
Workflow Clarity | The install → auth → connect → poll → handle clientAction → search → run sequence is present and the connection-state decision handling (READY / CLIENT_ACTION_REQUIRED / CONFIGURATION_ERROR) is genuinely useful, but the body says 'skip to **Step 2**' and no Step 2 is ever labeled (and '1b' has no '1a'), section ordering is jumbled, and there is no verification of action-run results or error handling after 'action run'. Matches anchor 3 ('Steps listed but validation gaps'); not 4, because the broken cross-references and missing run-result checkpoints are more than minor gaps. | 3 / 5 |
Progressive Disclosure | A single well-sectioned file with no bundle files and no nested references — navigation is possible — but at ~140 lines it exceeds the simple-skill exception, and organization gaps remain: a dangling entity list with no purpose, an empty 'Popular actions' heading, and broken step numbering. Matches anchor 3 ('Some structure but could be better organized'); not 4, because content that should be curated or split (entity overview, actual popular actions) is either missing or unintegrated. | 3 / 5 |
Total | 13 / 20 Passed |