Content
27%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The skill contains useful, actionable Membrane CLI instructions for connecting to 0codekit and running actions, but these are buried under an enormous, token-wasteful enumeration of hundreds of integrations that provides no actionable guidance. The opening statement 'I don't have enough information to do that' is counterproductive. The actionable sections (authentication, connection, action search/run, proxy) are decent but would benefit from better organization and the removal or externalization of the massive overview list.
Suggestions
Remove or move the massive 500+ line integration overview list to a separate reference file — it consumes enormous token budget without providing actionable guidance.
Remove the opening line 'I don't have enough information to do that. Can you tell me more about the app?' as it's confusing and anti-actionable in a skill file.
Add descriptions to the popular actions table entries instead of 'No description' — these should explain what each action does and what inputs are expected.
Structure the content as a clear end-to-end numbered workflow: 1) Install CLI → 2) Authenticate → 3) Connect → 4) Discover actions → 5) Run actions, with explicit validation/error handling at each step.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The massive 'Overview' section listing hundreds of integrations (spreadsheets, calendars, mailboxes, chat apps, databases, file types, programming languages, etc.) is extremely verbose and provides no actionable value. This consumes enormous token budget with a simple enumeration that Claude doesn't need and that provides no guidance on how to interact with any of these services. | 1 / 3 |
Actionability | The Membrane CLI commands for authentication, connection, action search, action execution, and proxy requests are concrete and copy-paste ready. However, the skill opens with 'I don't have enough information to do that' which is anti-actionable, and the massive overview list provides zero executable guidance. The popular actions table lacks descriptions, reducing actionability. | 2 / 3 |
Workflow Clarity | The connection workflow (ensure → check state → handle CLIENT_ACTION_REQUIRED → poll until READY) is reasonably well-sequenced with state-based branching. However, there's no explicit validation checkpoint after running actions, no error recovery guidance for failed action runs, and the overall flow from authentication through to action execution could be more clearly delineated as a numbered end-to-end workflow. | 2 / 3 |
Progressive Disclosure | The skill is a monolithic wall of text with the enormous overview list inline rather than in a separate reference file. There are no references to supporting files for the integration catalog, API details, or advanced usage. The content that should be split out (the 500+ line integration list) dominates the file and buries the actionable content. | 1 / 3 |
Total | 6 / 12 Passed |