Content
27%Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This skill suffers severely from a massive, likely hallucinated entity list (~150 items) that appears to be from a legal practice management platform rather than Adversus (a dialer/CRM for outbound sales). The Membrane CLI integration instructions are reasonably clear but generic across all Membrane-connected apps. The skill would benefit enormously from removing the entity list, adding Adversus-specific examples with real parameters, and restructuring for conciseness.
Suggestions
Remove the entire 150+ entity bullet list which appears hallucinated/incorrect for Adversus and wastes massive token budget — replace with a brief description of Adversus's actual data model (leads, campaigns, activities, etc.)
Add concrete, executable examples for the popular actions table, e.g., show a complete `membrane action run create-lead --input '{"name": "John Doe", "phone": "+1234567890"}'` with expected output
Fix the product description — Adversus is an outbound dialer/CRM platform, not a 'competitive intelligence platform' — this fundamental error undermines the skill's credibility
Extract reference material (proxy request flags table, connection state handling details) into separate bundle files and keep SKILL.md as a concise overview with links
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The massive bullet-point list of 150+ entity types (Case, Contact, Task, etc.) is overwhelmingly verbose and provides no actionable value. Many of these entities (e.g., Pet, Vehicle, Will, Trust, Power of Attorney, Healthcare Directive) appear to be from a legal practice management platform, not a competitive intelligence tool like Adversus. This is clearly boilerplate/hallucinated content that wastes enormous token budget. | 1 / 3 |
Actionability | The Membrane CLI commands are concrete and executable (install, login, connection ensure, action run, proxy requests), but the Adversus-specific guidance is weak. The 'Popular actions' table provides useful action names but no executable examples with real parameters or expected outputs. The entity list provides zero actionable guidance. | 2 / 3 |
Workflow Clarity | The connection setup workflow is reasonably well-sequenced with state handling (READY, CLIENT_ACTION_REQUIRED, CONFIGURATION_ERROR), but there are no validation checkpoints for the actual Adversus operations. The workflow for discovering and running actions is present but lacks error handling or verification steps after running actions. | 2 / 3 |
Progressive Disclosure | The content is a monolithic wall of text with no references to external files. The enormous entity list should either be removed entirely or placed in a separate reference file. There is no bundle structure to support progressive disclosure, and the inline content is poorly organized with the massive list dominating the document. | 1 / 3 |
Total | 6 / 12 Passed |