mcp-server-wazuh
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@mcp-server-wazuh show me recent high-severity vulnerability findings"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
mcp-server-wazuh
stdio MCP server for Wazuh 5. It reads the manager API on port 55000 (agents, groups, cluster, MITRE) and the indexer on port 9200 (events, findings, vulnerability state).
Wazuh 5 removed the 4.x manager paths /manager, /rules, /syscheck, /sca and /vulnerability. Existing public servers such as gbrigandi/mcp-server-wazuh still call those paths, so this package talks to the 5.0 API instead of forking them. Rule matches are findings in wazuh-findings-v5*, not wazuh-alerts-*.
Install
apt install mcp-server-wazuh
apt install mcprack-mcp-server-wazuhRelated MCP server: Wazuh MCP Server
Environment
Variable | Secret | Default | Meaning |
| no |
| Manager API base URL |
| yes | – | API user. Used once to obtain a JWT |
| yes | – | API password |
| no |
| Indexer base URL |
| yes | – | Indexer user |
| yes | – | Indexer password |
| no |
| Hide |
| no |
| Verify manager and indexer certificates |
| no |
| Seconds per request |
| no |
| Truncate larger tool output |
| no |
| Events data stream |
| no |
| Findings data stream |
| no |
| Vulnerability state index |
The JWT is refreshed in-process and is never printed. A 401 fetches a new token once.
Tools
Tool | Title | Annotation | Backend |
| API Info | read-only | manager |
| List Agents | read-only | manager |
| Get Agent | read-only | manager |
| Agents Summary | read-only | manager |
| List Groups | read-only | manager |
| Cluster Status | read-only | manager |
| List Cluster Nodes | read-only | manager |
| Search MITRE | read-only | manager |
| Search Events | read-only | indexer |
| Search Findings | read-only | indexer |
| Search Vulnerabilities | read-only | indexer |
| Restart Agents | destructive | manager, only when |
restart_agents interrupts agents. Ask before calling it. Wazuh 5 may reject the restart endpoint.
Development
PYTHONPATH=src python3 -m pytest
Available Tools
11 toolsget_agentGet AgentCRead-onlyIdempotent
Read one agent from the manager API.
Inputs: agent_id (required). Returns: result
get_agents_summaryAgents SummaryARead-onlyIdempotent
Read agent counts grouped by status.
Inputs: no named parameters. Returns: result
get_api_infoAPI InfoARead-onlyIdempotent
Read the Wazuh manager API root. Confirms the host and the running build.
Inputs: no named parameters. Returns: result
get_cluster_statusCluster StatusARead-onlyIdempotent
Read whether the manager cluster is enabled and healthy.
Inputs: no named parameters. Returns: result
list_agentsList AgentsCRead-onlyIdempotent
List agents from the Wazuh 5 manager API.
Inputs: limit, offset, search, status. Returns: result
list_cluster_nodesList Cluster NodesBRead-onlyIdempotent
List Wazuh cluster nodes.
Inputs: no named parameters. Returns: result
list_groupsList GroupsCRead-onlyIdempotent
List agent groups.
Inputs: limit, offset. Returns: result
search_eventsSearch EventsCRead-onlyIdempotent
Search the wazuh-events-v5 data stream on the indexer.
Inputs: limit, query, offset, agent_id. Returns: result
search_findingsSearch FindingsARead-onlyIdempotent
Search rule matches in the wazuh-findings-v5 data stream. Wazuh 5 calls these findings, not alerts.
Inputs: limit, query, offset, agent_id. Returns: result
search_mitreSearch MITREBRead-onlyIdempotent
List MITRE techniques known to the manager.
Inputs: limit, offset. Returns: result
search_vulnerabilitiesSearch VulnerabilitiesARead-onlyIdempotent
Search the wazuh-states-vulnerabilities index, newest detections first. Example query: vulnerability.severity:Critical
Inputs: limit, query, offset, agent_id. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 11 tools
Each tool maps to a distinct resource (agents, groups, cluster, MITRE, events, findings, vulnerabilities) and the descriptions explicitly clarify the trickiest pair, search_events vs search_findings. The main residual overlap is that events and findings are closely related data streams, and get_agents_summary vs list_agents require reading descriptions to separate counts from records.
Consistent snake_case verb_noun pattern throughout: get_*, list_*, and search_* all used predictably against the noun they operate on. The few multi-word nouns (get_agents_summary, get_cluster_status) still follow the same verb-first convention.
Eleven tools is well within the ideal 3-15 range and every tool covers a distinct slice of the Wazuh read surface. No redundant or obviously filler tools are present.
The set covers the main read workflows: manager info, agents, groups, cluster health/nodes, MITRE, events, findings, and vulnerabilities. Gaps are minor but real: list_groups has no companion get_group, there is no way to fetch a single finding/event/vulnerability record by ID, and rules/decoders or manager logs are absent.
Maintenance
Related MCP Connectors
Read-only access to an ET Ducky organization's agents, alerts, and fleet summary
Read-only CVE intelligence, remediation playbooks, and agent setup guides. Not a scanner.
Read-only MCP access to a documented IT fleet: state, changes, posture. 15 tools.
Read-only honeypot observations and evidence. Invite-only API key; request access on our website.
Related MCP Servers
- FlicenseAqualityCmaintenanceEnables querying a remote Wazuh deployment via MCP, covering Manager API (agents, rules, inventory, configuration) and Indexer (alerts, vulnerabilities) with 34 tools and reports, read-only by default.34-
- AlicenseNot gradedqualityCmaintenanceEnables AI assistants to converse in plain language with a Wazuh SIEM deployment, querying and aggregating alerts, hunting threats, triaging vulnerabilities, running compliance checks, and executing or rolling back active responses across multi-cluster environments. Works with both cloud and fully local, air-gapped LLM clients while enforcing scoped, audited access to the security tools.MIT
- AlicenseNot gradedqualityBmaintenanceExposes Wazuh SIEM data and actions—including agents, alerts, vulnerabilities, rules, decoders, active response, and Wazuh 5 Sigma rules, Detectors, and content promotion—as Model Context Protocol tools that LLM clients can call over HTTP.MIT
- AlicenseNot gradedqualityCmaintenanceEnables MCP clients to query and manipulate a Wazuh SIEM deployment through its manager, Indexer, and dashboard REST APIs, covering agents, alerts, vulnerabilities, rules, decoders, active response, and content promotion.MIT