Unofficial MCP server for HPE Aruba Networking Central
Single tool 'central_get_alerts' has a well-structured schema with proper enums, types, and descriptions. Description is detailed (250+ chars) and explains WHAT, WHEN, and HOW. Parameters are typed and constrained (enums for status/device_type/category, min/max for limit). However, output schema is not documented in the visible code, only mentioned as 'AlertEnvelope' without field definitions. Tool naming is clear (verb_noun pattern). No critical security issues detected. Pagination support (cursor/limit) is present. Linting rules (D, ANN) are strictly enforced in tools/, indicating high description/annotation discipline.
Retrieve a severity-sorted alert page for one site. Use after central_get_sites identifies the site; Active alerts are the default. Cross-field rule: filters combine, and cursor requires the identical query and original limit. Returns AlertEnvelope; response_format selects concise or detailed items. Replay next_cursor as cursor to continue.
Output schema (AlertEnvelope) is not documented. LLMs cannot infer what fields to expect or plan downstream calls. Rubric requires 'Document the output schema' for all tools.
No error handling guidance in tool description. If API returns 404 (site not found) or 401 (auth failed), the description does not tell the LLM what to do next or how to recover.
Tool description mentions 'Cross-field rule: filters combine, and cursor requires the identical query and original limit' but does not explain what happens if the rule is violated. LLMs need explicit guidance on parameter dependencies.
Only 1 tool registered. Server is incomplete for a Central monitoring use case, missing tools like central_get_sites (mentioned in description as prerequisite), central_get_devices, central_acknowledge_alert, etc. Limits agent capability.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | C | 67 | 2026-07-28+ | v2 |