Content
61%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 solid, command-driven integration skill: nearly every section gives executable bash, the connection lifecycle is handled with a sensible state decision tree, and auth edge cases (headless, re-auth) are addressed. Its main defects are a dangling "Step 2" reference that breaks the workflow's coherence, mild duplication and CLI-invocation inconsistency, and keeping borderline reference material inline in a file with no progressive-disclosure layer at all.
Suggestions
Fix the dangling navigation: the body says "skip to **Step 2**" twice but never defines Step 2 — either number the workflow steps explicitly or replace the reference with the actual section name (e.g. "skip to Searching for actions").
Remove the duplicated `action list` command ("Popular actions" repeats "Searching for actions") and standardize on one CLI invocation (`membrane` vs `npx @membranehq/cli`), explaining when each applies.
Trim the marketing sentence about employee engagement and consider moving the proxy flags table and clientAction schema detail into a reference file to keep SKILL.md as a lean overview.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient command-first prose, but there is unnecessary material: the marketing sentence "The software aims to improve employee engagement and drive organizational growth", a duplicated `action list` invocation in both "Searching for actions" and "Popular actions", and inconsistent CLI invocation (`membrane` vs `npx @membranehq/cli`). It matches 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the minor-trim-only bar of 4. | 3 / 5 |
Actionability | Concrete copy-paste bash covers the whole flow: install (`npm install -g @membranehq/cli@latest`), login including the headless `membrane login complete <code>` path, `connection ensure`, polling with `--wait`, `action run` with `--input`, and a proxy flags table. Minor gaps keep it from 5: no worked example with a real action id or query, and the CLI is invoked two different ways without explanation. | 4 / 5 |
Workflow Clarity | A real sequence exists (install → authenticate → connect → poll state → discover/run actions → proxy fallback) with a useful state decision tree (READY / CLIENT_ACTION_REQUIRED / CONFIGURATION_ERROR). But the document says "skip to **Step 2**" in two places while no Step 2 is ever defined, leaving the reader at a dead reference — sequence present with implicit checkpoints, matching anchor 3 rather than 4. | 3 / 5 |
Progressive Disclosure | No bundle files exist and the single body is well-sectioned with clear headers. At ~140 lines it is borderline: the proxy flags table and the clientAction schema walkthrough are candidates for a reference file, so structure is good with minor organization gaps rather than the exemplar split of anchor 5. | 4 / 5 |
Total | 14 / 20 Passed |