Production-grade Model Context Protocol server for Wazuh security platform using stdio transport
Server provides 8 tools with basic schemas and descriptions, but quality is inconsistent. All tools have descriptions (10-60 chars, mostly short), and input schemas are present with type definitions. However, descriptions are generic and lack LLM-optimized guidance on WHEN to use each tool or HOW to interpret results. Parameter descriptions are minimal (e.g., 'Filter by agent ID' without format/constraint details). No output schemas documented, LLMs cannot predict response structure. Error handling is absent from tool definitions. Tool naming is acceptable (verb_noun pattern) but some tools overlap in purpose (get_wazuh_alerts vs get_wazuh_alert_summary). No tool annotations (readOnlyHint, destructiveHint) despite all being read-only. Composition is weak: no pagination guidance, no chaining IDs documented, no recovery hints for common failures.
Analyze a security threat indicator using AI-powered analysis.
Retrieve information about Wazuh agents.
Get a summary of Wazuh alerts grouped by specified field.
Retrieve Wazuh security alerts with optional filtering.
Get Wazuh cluster health information.
Get summary of Wazuh rules and their effectiveness.
No output schemas documented. LLMs cannot predict response structure, field names, or types. Agents must guess what fields to extract for downstream calls.
Descriptions are too short (10-60 chars) and lack LLM-optimized guidance. Missing WHEN to use each tool, WHAT it returns, and HOW to interpret results. E.g., 'Retrieve Wazuh security alerts with optional filtering' does not explain when to call this vs get_wazuh_alert_summary.
Inferred effective spec: <=2025-11-25.
| Scored | Grade | Overall | Spec posture | Rubric |
|---|---|---|---|---|
| 2026-09-23 | D | 51 | <=2025-11-25 | v2 |
Get list of currently running/active Wazuh agents.
Retrieve vulnerability information from Wazuh.
Parameter descriptions lack format/constraint details. E.g., 'Filter by alert level (e.g., "12", "10+")' uses examples instead of formal constraints. LLMs may hallucinate invalid values like '5' or 'critical'.
No pagination guidance. Tools like get_wazuh_alerts accept 'limit' (1-1000) but no offset/cursor or total_count in response. Large result sets will blow context window.
No error handling guidance in tool definitions. LLMs have no recovery path if a call fails (e.g., 'agent_id not found', should they retry, call get_wazuh_agents first, or ask the user?).