Content
67%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 provides strong, executable CLI guidance with a genuinely thorough connection lifecycle workflow including error-state feedback loops. Its main weaknesses are token waste in the placeholder action table, a dangling 'Step 2' reference that breaks the documented sequence, and no progressive disclosure of the bulk catalogue into reference files.
Suggestions
Replace the 20-row table whose Description column is uniformly 'No description' with brief real descriptions (or drop the column), since it burns tokens conveying only action keys.
Fix the dangling 'skip to Step 2' reference and renumber the connection sections (e.g. '1a. Ensure a connection' / '1b. Wait for the connection') so the documented flow is self-consistent.
Delete filler lines such as 'Use action names and parameters as needed.' and add one worked example of `membrane action run` with a realistic input payload.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly lean commands, but the 20-row Popular actions table wastes tokens with 'No description' on every row, and filler lines like 'Use action names and parameters as needed.' add nothing. It is below 4 because of this noticeable padding. | 3 / 5 |
Actionability | Concrete copy-paste commands cover install, login (including headless completion), connection setup, polling, action discovery/execution, and proxy usage with a flag table. Minor gaps — placeholders like <actionId> are never instantiated and there is no worked example with real input JSON — keep it below 5. | 4 / 5 |
Workflow Clarity | The setup sequence (install → login → connection ensure → poll until READY → search actions → run) is clear, with error-state handling (CLIENT_ACTION_REQUIRED, CONFIGURATION_ERROR/SETUP_FAILED) acting as validation checkpoints. Below 5 because the text references 'Step 2' which is never defined, and section '1b' appears without a corresponding '1'/'1a', leaving minor sequence gaps. | 4 / 5 |
Progressive Disclosure | There are no bundle files, and the single SKILL.md (~165 lines) is well organized into install, authentication, connection, actions, and best-practices sections with tables. Minor gaps: the bulk action catalogue and proxy flag reference would sit naturally in separate reference files. | 4 / 5 |
Total | 15 / 20 Passed |