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 a solid, command-driven integration guide: real CLI invocations, a well-documented connection-state/polling flow, and useful best practices. Its weaknesses are broken workflow numbering (references to a nonexistent 'Step 2' and an orphan '1b'), inlined boilerplate that inflates token cost, and no progressive-disclosure structure for the reference material. Fixing the step labels and trimming the generic sections would lift it considerably.
Suggestions
Fix the broken sequencing: either label the sections as numbered steps so 'skip to Step 2' resolves, or rewrite as plain prose/flow without step references; also resolve the orphan '1b' heading.
Trim content Claude doesn't need: the opening explanation of what Cognito is, the 'Use action names and parameters as needed' filler line, and the generic 'Agent Types' paragraph.
Add one realistic end-to-end example (e.g. list a user pool's users via action search → run with actual --input JSON) and move the generic Membrane CLI/auth and proxy-flag material to a shared reference file if the platform allows bundling.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Mostly efficient command-driven prose, but it spends tokens on things Claude already knows or doesn't need: the opening paragraph explaining what Cognito is, the filler line "Use action names and parameters as needed", and the generic "Agent Types" aside. This matches 'Mostly efficient but includes some unnecessary explanation or could be tightened'; it is not 4 because several such passages could be trimmed outright. | 3 / 5 |
Actionability | Nearly every section gives an executable command with real flags (`npm install -g @membranehq/cli@latest`, `membrane connection ensure "https://cognitohq.com/" --json`, the proxy flag table). Minor gaps keep it below 5: `membrane login --tenant` shows no value, and action examples stop at `--input '{"key": "value"}'` with no realistic Cognito example. | 4 / 5 |
Workflow Clarity | The overall sequence (install → login → connection → search actions → run) is present, but the step references are broken: "skip to **Step 2**" points to a section that is never labeled Step 2, and "#### 1b" appears with no 1a or numbered steps. Sequence present but checkpoints/structure are implicit, matching anchor 3 rather than 4. | 3 / 5 |
Progressive Disclosure | There are no bundle files (no references/, scripts/, or assets/ directories) and no external file references, so everything lives in one ~140-line SKILL.md. Headers provide some structure, but generic Membrane material (CLI auth flow, agent types, proxy flag table) is inlined where a per-app overview belongs, matching 'Some structure but could be better organized'. | 3 / 5 |
Total | 13 / 20 Passed |