Content
75%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 strong, executable operational guide: concrete CLI commands cover setup, connection state handling, action discovery/execution, and a proxy fallback. Weaknesses are a padded intro paragraph, a dangling "Step 2" reference in the workflow, and no out-of-line reference files for the lengthy action catalog.
Suggestions
Delete or trim the opening platform-summary paragraph — Code Climate's purpose and user base are background Claude does not need to operate the CLI.
Fix the broken workflow numbering: either label the sections Step 1/1b/2 consistently or replace the "skip to Step 2" references with descriptive section names.
Add a brief validation note for destructive actions (e.g. confirm repository name and verify with get-repository before running delete-repository).
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly commands and tables, but the intro paragraph ("Code Climate is a platform that helps software engineering teams improve code quality... It's used by developers, QA engineers, and engineering managers") pads with who-uses-it background Claude does not need, and phrases like "This will burn less tokens" add fluff. This fits 'mostly efficient with some unnecessary explanation' rather than the more heavily padded anchor 2. | 3 / 5 |
Actionability | Commands are copy-paste ready throughout: install (npm install -g @membranehq/cli@latest), both interactive and headless auth flows, connection ensure with polling, action list/run with a concrete --input JSON example, and a full proxy flag table. Minor placeholders (CONNECTION_ID) and one npx-vs-global inconsistency do not undermine coverage of the common cases. | 5 / 5 |
Workflow Clarity | A clear install → login → connect → poll → search → run sequence exists with an explicit state machine (READY / CLIENT_ACTION_REQUIRED / CONFIGURATION_ERROR) and a retry loop after the user completes an action. Gaps: "skip to Step 2" references a Step 2 that is never labeled, the "1b" heading has no companion numbering, and the destructive delete-repository action gets no validation guidance. | 4 / 5 |
Progressive Disclosure | Sections are well organized (install, auth, connect, search, run, proxy, best practices) with no nested references, but there are no bundle files at all — the 20-row action catalog is inlined in a ~158-line SKILL.md. The table doubles as the discovery surface so inline is defensible, fitting 'good structure; most content appropriately placed; minor organization gaps' rather than the well-split reference-file anchor 5. | 4 / 5 |
Total | 16 / 20 Passed |