Content
65%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 skill is highly actionable with executable CLI commands and a solid connection-state feedback loop, but it bloats the context window with an unnecessary marketing intro and a large inline action reference while omitting validation steps for destructive and batch operations.
Suggestions
Trim the opening paragraph ("Moosend is an email marketing automation platform. It's used by businesses of all sizes...") and the "burn less tokens" phrasing; Claude does not need the market positioning.
Add an explicit validation/confirmation checkpoint before destructive and batch actions (e.g., confirm the mailing list ID with the user before remove-subscriber, or run a get-* detail call to verify the target before delete-campaign/delete-mailing-list).
Move the popular-actions table into a separate references file (e.g. ACTIONS.md) and keep SKILL.md as an overview that links to it, so the action catalog does not consume the context window by default.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient with concrete commands, but opens with marketing-style padding ("It's used by businesses of all sizes to create, send, and track email campaigns...") and includes filler like "burn less tokens and make communication more secure" that assumes less competence than warranted. | 3 / 5 |
Actionability | Commands are copy-paste ready with real flags and package names (e.g. `npm install -g @membranehq/cli@latest`, `membrane action run <actionId> --connectionId=CONNECTION_ID --input '{"key": "value"}' --json`) and cover the common cases of login, connect, search, run, and proxy. | 5 / 5 |
Workflow Clarity | The connection flow has a genuine feedback loop (poll until READY, handle CLIENT_ACTION_REQUIRED, re-poll), but destructive/batch actions like delete-campaign, remove-subscriber, and add-multiple-subscribers are executed with no validation or confirmation step, capping this dimension at 3 per the rubric. | 3 / 5 |
Progressive Disclosure | Section structure is clear, but the 20-row popular-actions reference table is inlined content that belongs in a separate reference file, and there are no bundle files or external file references to offload the bulk. | 3 / 5 |
Total | 14 / 20 Passed |