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.
The body is dense and highly executable — exact CLI commands, flag tables, and a connection state machine with explicit re-polling — but it opens with an unnecessary product explainer, never demonstrates a real end-to-end action run, and offers no verification around the destructive operations it exposes. Numbering inconsistencies ("Step 2" and "1b" without parents) slightly mar an otherwise navigable structure.
Suggestions
Cut the opening paragraph explaining what ClickUp is and start directly with the Membrane CLI workflow.
Add verification guidance for destructive operations: confirm the task/list and its contents before running delete-task/delete-list, and check the run output before reporting success.
Fix the step numbering — "skip to **Step 2**" points to a step that is never labeled, and "#### 1b" has no "1a" — either number the workflow steps consistently or remove the numbers.
Show one worked example with real values (e.g. create-task with an actual input payload) and clarify whether actions are invoked by key (create-task) or by the id returned from action list.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The opening paragraph ("ClickUp is a project management platform used by individuals and teams... from small businesses to large enterprises") explains a product Claude already knows, matching anchor 3 ("Mostly efficient but includes some unnecessary explanation or could be tightened"); not 4 because a full padding paragraph exceeds "minor instances that could be trimmed", and not 2 since the bulk is dense command/flag/state-machine content with no library-comparison style padding. | 3 / 5 |
Actionability | Guidance is copy-paste ready throughout: "npm install -g @membranehq/cli@latest", "membrane connection ensure \"https://app.clickup.com/\" --json", "membrane action run <actionId> --connectionId=CONNECTION_ID --input '{\"key\": \"value\"}' --json", plus flag tables. Matches anchor 4 ("Mostly executable guidance... minor gaps"): the Popular actions table lists "Key" values (e.g. create-task) while the run command takes <actionId>, leaving the key-vs-id question unresolved and no worked example with real values; not 5 because common cases are never demonstrated end-to-end. | 4 / 5 |
Workflow Clarity | A clear sequence exists (install → login → connection ensure → poll state → action list → action run) with a real feedback loop for connection readiness ("poll again with `membrane connection get <id> --json` to check if the state moved to READY"). However, destructive operations in the actions table (Delete Task, Delete List) have no verification steps, which caps the score at 3 per the destructive/batch rule; additionally "skip to **Step 2**" references a step that is never labeled and "#### 1b" appears with no "1a". Not 2 because the connection state machine and re-poll loop are well defined. | 3 / 5 |
Progressive Disclosure | The single-file body is well organized with clear sections (Install, Authentication, Connecting, Searching, Popular actions, Running actions, Proxy requests, Best practices), and no bundle files exist so nothing is mis-filed or nested. Matches anchor 4 ("Good structure; most content is appropriately placed... minor organization gaps") — the 20-row actions catalog and clientAction schema detail could live in a reference file; not 5 because at ~175 lines (well past the 50-line simple-skill case) the monolithic placement leaves minor organization gaps. | 4 / 5 |
Total | 14 / 20 Passed |