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 delivers strong, executable CLI guidance with a genuinely useful connection state-machine, but is undermined by marketing filler, an empty 'Popular actions' section, and broken step numbering ('skip to Step 2' with no Step 2). No progressive disclosure: everything, including reference-grade detail, is inlined in one file.
Suggestions
Fix the workflow numbering: label the steps explicitly (1. Install, 2. Authenticate, 3. Connect, 4. Discover actions, 5. Run actions) so 'skip to Step 2' resolves, and add a brief 'if action run fails, check the error field' checkpoint.
Move the proxy flags table and the connection clientAction state-machine detail into a references/ file (e.g., REFERENCES.md), keeping SKILL.md a concise overview with clearly signaled one-level-deep links.
Either populate 'Popular actions' with 2-3 worked Cludo examples (e.g., search with a real inputSchema payload) or remove the section, and cut the marketing intro ('used by businesses of all sizes...') and the content-free 'Use action names and parameters as needed.' line.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The command documentation is tight, but marketing filler ('used by businesses of all sizes to improve their website's search functionality and user experience') and content-free lines ('Use action names and parameters as needed.', the bare Search/Result/Settings/Account bullet tree) pad the body. It is not 4 because there are several unnecessary sections, not just minor trimmable instances. | 3 / 5 |
Actionability | Nearly all guidance is copy-paste executable: install ('npm install -g @membranehq/cli@latest'), login with headless fallback, 'membrane connection ensure', action list/run with --input JSON, and a proxy flags table. It is not 5 because the 'Popular actions' header promises examples but delivers none, so common Cludo cases (search, get results) lack a worked example. | 4 / 5 |
Workflow Clarity | A connection state machine with polling feedback loops is present (READY / BUILDING / CLIENT_ACTION_REQUIRED with 'poll again... to check if the state moved to READY'), but 'skip to Step 2' references a step that is never labeled — numbering jumps to '#### 1b' with no visible Step 1 or 2 — and the action-run path has no error-handling guidance. It is not 4 because the dangling step reference is a genuine sequence gap, not a minor one. | 3 / 5 |
Progressive Disclosure | The ~140-line body is monolithic with no bundle files; detail like the proxy flag table and the connection clientAction state machine is inlined where a reference file would fit, and internal step labels are inconsistent. It is not 4 because content that clearly belongs in separate files is inline and there are no clearly signaled references at all. | 3 / 5 |
Total | 13 / 20 Passed |