Content
68%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 highly actionable, with executable CLI commands for every phase of connecting to and calling Cloze through Membrane, and it is organized into clear sections. Its main weaknesses are a dangling 'Step 2' cross-reference, no validation guidance for the destructive delete actions it advertises, and a few trimmable explanatory passages.
Suggestions
Fix the dangling 'Step 2' references: either add the actual Step 2 section (e.g., 'Step 2: Search for actions') or rewrite as 'skip to Searching for actions' so the reader is never pointed at a nonexistent step.
Add a validation checkpoint before destructive actions (delete-project, delete-company, delete-person) — e.g., get the record first with get-company/get-person and confirm with the user before running the delete — to provide a feedback loop for irreversible operations.
Trim the unnecessary explanation: drop 'Use action names and parameters as needed.', tighten the intro paragraph about what Cloze is, and clarify the vague 'Agent Types' note with the actual --clientName values to use.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient commands and tables, but it includes unnecessary explanation that could be tightened: an intro paragraph describing what Cloze is and who uses it, the filler line 'Use action names and parameters as needed.', and a vague 'Agent Types' note ('Those will be used to adjust tooling...'). This fits anchor 3 ('Mostly efficient but includes some unnecessary explanation or could be tightened') better than anchor 4, since there are several such spots, not just minor trimmable instances. | 3 / 5 |
Actionability | Every command is concrete and copy-paste ready: npm install, membrane login with headless completion flow, connection ensure with --wait polling, action list/run with a JSON --input example, and a complete proxy flag table. This matches anchor 5 ('Fully executable; copy-paste ready code or commands; specific examples cover the common cases'). | 5 / 5 |
Workflow Clarity | The install → login → connect → search → run sequence is present with state-based branching (READY / CLIENT_ACTION_REQUIRED / CONFIGURATION_ERROR), but the text says 'skip to **Step 2**' twice and no Step 2 exists, and the skill advertises destructive actions (delete-project, delete-company, delete-person) with no validation or verification guidance. The destructive-operations cap applies, so this cannot exceed anchor 3 ('Steps listed but validation gaps; sequence present but checkpoints missing or implicit'). | 3 / 5 |
Progressive Disclosure | There are no bundle files; the single ~160-line body is organized with clear section headers (Install, Authentication, Connecting, Searching, Popular actions, Running, Proxy, Best practices), and the 18-row action table is compact enough to serve discovery inline rather than needing a separate reference file. This fits anchor 4 ('Good structure; most content is appropriately placed; minor organization gaps') — not anchor 3, since there is no large block that clearly belongs in a separate file, and not anchor 5, which expects well-signaled one-level-deep references. | 4 / 5 |
Total | 15 / 20 Passed |