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 delivers strong executable guidance: exact Membrane CLI commands for install, auth, connection setup with state handling, action discovery/execution, and proxied raw API calls. Weaknesses are a padded intro and feature list, dangling references to a nonexistent "Step 2", and no validation guidance for the destructive delete actions the skill advertises.
Suggestions
Cut the "ChMeetings Overview" bullet list and the intro paragraph about who uses the software — keep only the one-line description and the API docs link.
Fix the navigation references: either number the steps (Step 1: connect, Step 2: run actions) or replace the two "skip to **Step 2**" references with concrete section names like "Searching for actions".
Add a short validation note for destructive operations, e.g. "Before running delete-person or delete-family, fetch the record with get-person/get-family and confirm the ID with the user".
Include one worked example of running an action with real values, such as `membrane action run list-people --connectionId=<id> --input '{"page": 1}' --json`.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly lean commands and tables, but the two-sentence intro about who uses ChMeetings and the 24-item "ChMeetings Overview" feature bullet list (Meetings through Support) add ~30 lines of non-actionable padding. This fits 'Mostly efficient but includes some unnecessary explanation or could be tightened'; not a 4 because the feature list and intro prose are clearly trimmable. | 3 / 5 |
Actionability | Nearly every step has a copy-paste-ready command ("npm install -g @membranehq/cli@latest", "membrane connection ensure \"https://www.chmeetings.com/\" --json", "membrane action list --connectionId=... --intent ... --json") plus a flags table for proxy requests. Not a 5 because the parameter example is a generic '{"key": "value"}' and there is no worked end-to-end example with real values (e.g. running list-people against a connection id); not a 3 because all guidance is executable, not pseudocode. | 4 / 5 |
Workflow Clarity | The setup sequence is present and state-aware (poll until READY, branch on CLIENT_ACTION_REQUIRED / CONFIGURATION_ERROR, re-poll after user action), but the text twice says "skip to **Step 2**" while no Step 2 (or numbered Step 1) exists, and the skill exposes destructive actions (delete-family, delete-person) with no validation/verification guidance — the rubric caps destructive-operation workflows without validation at 3. Not a 2 because the sequence and state checkpoints are otherwise concrete and coherent. | 3 / 5 |
Progressive Disclosure | This is a single-file skill with no bundle files, and its section structure (Install, Authentication, Connecting, Searching, Popular actions, Running, Proxy, Best practices) is clear and well-ordered for navigation at its ~180-line size. Not a 5 because the popular-actions table and detailed connection-state handling are borderline content that could live in a reference file, and the feature-list section is structureless padding; not a 3 because nothing is buried and no content is misplaced. | 4 / 5 |
Total | 14 / 20 Passed |