wazuh-mcp-server
This MCP server provides AI assistants with 28 tools to query, analyze, and act on Wazuh SIEM/XDR data across 9 security domains.
Alert Triage: List, fetch, and summarize Wazuh alerts by severity, rule, MITRE technique, agent, or free-text search.
Threat Hunting: Search raw events, query File Integrity Monitoring (FIM) records, investigate vulnerabilities by CVE/severity, and explore MITRE ATT&CK techniques.
Compliance Auditing: Retrieve SCA status/checks, generate fleet-wide compliance reports (CIS, PCI DSS, NIST, GDPR, HIPAA).
Fleet Management: List/inspect agents and groups, get fleet health, status counts, OS/version breakdowns, and stale agents.
Threat Intelligence: List and read CDB lists (IP blocklists, IOC databases, whitelists).
Manager & Cluster Ops: Get daemon stats, manager logs, cluster status, per-node stats, and search detection rules by framework/MITRE.
Security Analysis: Build MITRE/NIST/PCI/GDPR/HIPAA rules coverage maps, CVE vulnerability heatmaps, and incident timelines from an alert.
Incident Response (safety-gated): Trigger active-response commands (firewall-drop, host-deny, restart) or run agent commands via a two-step confirmation + token flow with audit logging.
Role-Based Access: Disabled by default; when enabled, restricts tools to viewer/analyst/admin/soc roles.
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., "@wazuh-mcp-server Show me all critical alerts from today"
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.
π SB SIEM MCP
Note: Independent, third-party project β not affiliated with or endorsed by Wazuh Inc. Actively developed and tested against live Wazuh instances; review and test before production deployment.
28 MCP tools. 9 domains. Tested on Wazuh 4.14.5 against live instances. AI-powered security operations for Wazuh SIEM/XDR.
"Show me all critical alerts in the last 6 hours, cross-reference with MITRE ATT&CK, and check if any affected hosts have unpatched CVEs."
One prompt. Your AI assistant queries 7,514 alerts, checks 5,038 FIM records, scans 12 CVEs, cross-references 750 MITRE techniques, audits CIS compliance, and triggers incident response β all through your Wazuh infrastructure.
How It Works (30 seconds)
You already have Wazuh running somewhere. The MCP server is a local process that your AI client spawns as a child β just like a language server or linter.
Your Machine Your Wazuh Server
ββββββββββββββββββββββ ββββββββββββββββββββ
β Zed / Claude β β β
β β β β Wazuh API β
β βΌ β β :55000 β
β python -m ββββββββHTTPSβββββββΆβ β
β wazuh_mcp.server β β Wazuh Indexer β
β (child process) ββββββββHTTPSβββββββΆβ :9200 β
ββββββββββββββββββββββ ββββββββββββββββββββNo Docker required. No containers. No agents to install. Just point it at your existing Wazuh and start asking questions in natural language.
Related MCP server: Wazuh MCP Server
π‘οΈ Security Features (Defense in Depth)
Input validation β Shell metacharacter blocking, regex for agent IDs, IPs, CVEs, MITRE IDs
Rate limiting β Token-bucket: 30/60s for read tools, 5/120s for destructive
Output sanitization β Redacts AWS keys, JWT tokens, SSH keys, API keys, passwords from LLM-bound data
Audit logging β Append-only JSONL trail for all destructive actions
Confirmation gate β Two-step
confirm=True+ expiring token for active response toolsRBAC β 4 built-in roles:
viewer,analyst,admin,socwith hierarchical accessDependabot + pip-audit + CodeQL β Automated dependency scanning on every push + weekly schedule
Non-root Docker β Production container runs as unprivileged
wazuhmcpuserTLS support β Verify server certificates, or disable for local dev via
WAZUH_INSECUREPrometheus metrics β 7 metrics defined and served on
:9090/metricsin SSE mode, but no tool records to them yetOpenAPI 3.0 spec β static
openapi.jsongenerated fromopenapi.py(22 of 28 tools); not served by the server
π Architecture
ββββββββββββββββββββββββββββ ββββββββββββββββββββββββββββ
β Your AI Client β β Wazuh Infrastructure β
β (Zed / Claude / Cursor) β β β
β β β β Wazuh API :55000 β
β βΌ β β ββ Agents, Groups β
β ββββββββββββββββββββ β β ββ SCA, FIM, MITRE β
β β MCP Server βββββHTTPSβββββΆβ ββ Manager, Cluster β
β β 28 tools β β β ββ Active Response β
β β β β β β
β β WazuhClient βββββΌββββHTTPSβββββΆβ Wazuh Indexer :9200 β
β β IndexerClient ββββ€ β β ββ Alerts (7,514+) β
β β RateLimiter β β β ββ Vulnerabilities β
β β Sanitizer β β β ββ Events, Rules β
β β RBACEnforcer β β β β
β ββββββββββββββββββββ β ββββββββββββββββββββββββββββ
β β β
β :9090/metrics β
β :8000/sse (SSE mode) β
ββββββββββββββββββββββββββββThe MCP server talks to both the Wazuh REST API (port 55000, for management) and the Wazuh Indexer (port 9200, for alerts/vulnerabilities). In Wazuh 4.x/5.x, alerts and vulnerabilities are indexer-only β not available via the REST API. The server's IndexerClient handles this transparently.
π What This Does
Workflow | Example Prompt | Tools Used |
Alert Triage | "Summarize today's alerts by severity and MITRE technique" |
|
Threat Hunting | "Search for IOC 10.0.0.50 across all events and FIM records" |
|
Compliance Audit | "Show me all agents failing CIS benchmark checks" |
|
Rules Coverage | "What's my NIST 800-53 detection coverage?" |
|
Vulnerability Mgmt | "Which systems have critical unpatched CVEs?" |
|
Incident Timeline | "Reconstruct what happened around alert #45821" |
|
Fleet Management | "List disconnected agents and their groups" |
|
Threat Intel | "Show me the CDB blocklists and MITRE techniques for T1059" |
|
Incident Response β οΈ | "Block IP 203.0.113.55 on all web servers" |
|
π¦ Installation
pip (from PyPI β coming soon)
pip install sb-siem-mcpFrom source
git clone https://github.com/Sbharadwaj05/sb-siem-mcp.git
cd sb-siem-mcp
pip install -e ".[dev]"Docker (oneβcommand demo β spins up Wazuh + MCP for testing)
β οΈ This bundles a full Wazuh stack for quick demos. In production, you already have Wazuh running β just use the pip install above and point to your existing Wazuh.
π The MCP endpoint has no built-in client authentication. The demo server binds to
localhost:8000β keep it local. For remote access, place it behind a reverse proxy with auth. See Production Hardening below.
git clone https://github.com/Sbharadwaj05/sb-siem-mcp.git
cd sb-siem-mcp
docker compose up -d
# Wazuh Dashboard: https://localhost:443
# MCP SSE endpoint: http://localhost:8000/sseConfiguration
Create a .env file:
# Required
WAZUH_API_URL=https://your-wazuh-manager:55000
WAZUH_USERNAME=wazuh-wui
WAZUH_PASSWORD=your-api-password
# Required for alerts, vulnerabilities, rules (Wazuh 4.x/5.x)
WAZUH_INDEXER_URL=https://your-wazuh-manager:9200
WAZUH_INDEXER_USER=admin
WAZUH_INDEXER_PASS=your-indexer-password
# Optional
WAZUH_INSECURE=true # Skip TLS verification (dev only)
WAZUH_RBAC_ROLE=analyst # Restrict tools by role
WAZUH_RATE_LIMIT_TOKENS=30 # Rate limit burst
WAZUH_RATE_LIMIT_PERIOD=60 # Rate limit windowImportant: The Wazuh Indexer (port 9200) must be accessible from the MCP server. By default it only listens on
localhost. See Troubleshooting for the one-line fix.
Claude Desktop / Zed / Cursor
{
"mcpServers": {
"wazuh": {
"command": "python",
"args": ["-m", "wazuh_mcp.server"],
"cwd": "/path/to/sb-siem-mcp/src",
"env": {
"WAZUH_API_URL": "https://192.168.56.102:55000",
"WAZUH_USERNAME": "wazuh-wui",
"WAZUH_PASSWORD": "your-api-password",
"WAZUH_INSECURE": "true",
"WAZUH_INDEXER_PASS": "your-indexer-password"
}
}
}
}π§ Complete Tool Reference (28 tools, 9 domains)
π Alerts & Triage (3)
Tool | Description | Data Source |
| Query alerts by severity, agent, rule ID, MITRE, search | Wazuh Indexer |
| Fetch single alert by ID with full context | Wazuh Indexer |
| Aggregated: severity distribution, top rules/IPs, MITRE coverage | Wazuh Indexer |
π Threat Hunting (4)
Tool | Description | Data Source |
| Submit raw events for Wazuh parsing/analysis | Wazuh API |
| File Integrity Monitoring β file changes, additions, deletions | Wazuh API |
| CVE inventory for one agent or the whole fleet, filterable by CVE and severity | Wazuh Indexer |
| MITRE ATT&CK techniques, tactics, mitigations, groups | Wazuh API |
π Compliance (3)
Tool | Description | Data Source |
| SCA policy scores per agent (CIS, PCI DSS, NIST, GDPR) | Wazuh API |
| Per-check pass/fail detail with rationales and remediation | Wazuh API |
| Fleet-wide compliance aggregation across all agents | Wazuh API |
π₯οΈ Agents & Groups (6)
Tool | Description | Data Source |
| List agents with status, OS, version, search, pagination | Wazuh API |
| Deep-dive on single agent: config, modules, groups | Wazuh API |
| Fleet health: status counts, OS breakdown, stale agents | Wazuh API |
| List agent groups with counts and checksums | Wazuh API |
| Group details, configuration, member counts | Wazuh API |
| All agents in a specific group | Wazuh API |
π CDB Lists (2)
Tool | Description | Data Source |
| List CDB threat-intel files (IP blocklists, IOC databases) | Wazuh API |
| Read contents of a CDB list file | Wazuh API |
βοΈ Manager & Cluster (5)
Tool | Description | Data Source |
| Daemon statistics (EPS, queues, processed events) | Wazuh API |
| Manager log retrieval with category and search filters | Wazuh API |
| Cluster health: enabled/running state | Wazuh API |
| Per-node daemon stats (falls back to manager stats for single-node) | Wazuh API |
| Search rules by framework/MITRE (falls back to indexer on 4.14.x bug) | Wazuh API / Indexer |
π Security Analysis (3)
Tool | Description | Data Source |
| MITRE/NIST/PCI/GDPR/HIPAA coverage matrix vs your rules | Wazuh Indexer |
| Risk-scored CVE heatmap across all agents | Wazuh Indexer |
| Auto-generated chronological attack timeline from an alert | Wazuh Indexer |
β οΈ Incident Response (2)
Tool | Description | Data Source |
| Trigger firewall-drop, host-deny, restart-wazuh (with confirmation gate) | Wazuh API |
| Execute command on remote agent (with confirmation gate) | Wazuh API |
π SAFETY: Destructive tools require two-step
confirm=True+ one-time expiring token. A misconfigured LLM cannot silently block IPs or quarantine hosts. All destructive actions are recorded in an append-only audit log.
π₯οΈ Observability
Prometheus Metrics (:9090/metrics)
Metric | Type | Description |
| Counter | Tool invocations by name + status (success/error) |
| Histogram | P50/P95/P99 latency per tool |
| Counter | Rate-limit rejections per tool |
| Gauge | Wazuh API connectivity (1=up, 0=down) |
| Counter | Destructive actions logged |
| Gauge | In-flight tool calls |
| Counter | Errors by tool + error type |
These are defined in src/wazuh_mcp/metrics.py and served in SSE mode on
WAZUH_MCP_HOST, but no tool calls the record_* helpers yet, so the values
stay at zero.
OpenAPI spec (openapi.json)
A static OpenAPI 3.0 description generated by python -m wazuh_mcp.openapi.
It covers 22 of the 28 tools and is not served by the MCP server.
Audit Log (~/.wazuh-mcp/audit.jsonl)
Append-only JSON Lines. One entry per destructive action. Never truncated. Thread-safe.
π RBAC
Four built-in roles with hierarchical, cumulative access:
Role | Access | Tools |
| Read-only | Alerts, agents, compliance, rules |
| + Investigation | All viewer + hunting, MITRE, CDB lists, analysis |
| + Administration | All analyst + manager stats, logs, cluster |
| + Response β οΈ | All admin + active response, agent commands |
export WAZUH_RBAC_ROLE=analyst
# Or custom policy: WAZUH_RBAC_POLICY=/path/to/rbac.jsonπ Project Structure
sb-siem-mcp/
βββ src/wazuh_mcp/
β βββ server.py # FastMCP entry point (stdio + SSE transport)
β βββ client.py # Wazuh REST API client (JWT, Basic Auth, fallback)
β βββ indexer.py # Wazuh Indexer / OpenSearch client (alerts, vulns)
β βββ rbac.py # Role-Based Access Control (4 roles, custom policies)
β βββ audit.py # Immutable audit logging (JSONL, append-only)
β βββ sanitizer.py # Output sanitization (credential redaction)
β βββ rate_limiter.py # Token-bucket per-tool rate limiting
β βββ validators.py # Input validation (regex, shell metacharacter blocking)
β βββ metrics.py # Prometheus metric definitions (not yet recorded by tools)
β βββ openapi.py # OpenAPI 3.0 spec generator (not served)
β βββ output.py # Token-efficient field selection (5 modes)
β βββ utils.py # JSON formatters, pagination helpers
β βββ tools/ # 9 tool modules, 28 MCP tools
β βββ alerts.py # 3 tools: list, get, summary
β βββ hunting.py # 4 tools: events, fim, vulns, mitre
β βββ compliance.py # 3 tools: sca status, checks, report
β βββ agents.py # 3 tools: list, get, health
β βββ groups.py # 3 tools: list, get, group agents
β βββ lists.py # 2 tools: list cdb, get cdb
β βββ manager.py # 5 tools: stats, logs, cluster, node, rules
β βββ analysis.py # 3 tools: coverage, heatmap, timeline
β βββ response.py # 2 tools: active response, agent command (safety-gated)
βββ tests/ # pytest suite: safety layer, regressions, client, versions
βββ docs/ # SECURITY, DEVELOPMENT, ADVANCED_FEATURES, TROUBLESHOOTING
βββ scripts/setup.sh # One-command Wazuh + MCP dev environment
βββ docker-compose.yml # Wazuh 4.14.5 demo stack + MCP server (see Known Limits)
βββ Dockerfile # Multi-stage production build (non-root)
βββ openapi.json # Generated OpenAPI 3.0 specification (24 paths)
βββ .github/workflows/ # CI (test matrix), Release, Security Scan
βββ CHANGELOG.md
βββ README.mdπ Quick Start
# 1. Clone
git clone https://github.com/Sbharadwaj05/sb-siem-mcp.git
cd sb-siem-mcp
# 2. Configure
cp .env.example .env
# Edit .env with your Wazuh API + Indexer credentials
# 3. Install
pip install -e ".[dev]"
# 4. Verify connectivity
python -c "
from wazuh_mcp.client import WazuhClient
import asyncio
async def t():
c = WazuhClient(insecure=True)
print('Agents:', (await c.list_agents(limit=1)).get('total_affected_items','?'))
print('Alerts:', (await c.list_alerts(limit=1)).get('total_affected_items','?'))
await c.close()
asyncio.run(t())
"
# 5. Connect to AI client
# Copy claude_desktop_config.json.example into your MCP configβ οΈ Known Limits
This section describes what the code does today. Each item names the code or test that shows it. Where this section disagrees with a feature description elsewhere in this README, this section is correct.
Confirmation gate
The gate stops a destructive action from running in a single call. The confirmation token is returned to the model in the tool output, so the gate on its own does not prove that a human approved. Human approval depends on your MCP client's tool-approval prompt; keep it on for
wazuh_run_active_responseandwazuh_agent_command. (src/wazuh_mcp/tools/response.py)A token executes only the action it was issued for: the same tool, agent, command and arguments. Any other use is refused and consumes the token. (
TestConfirmationGateandTestConfirmationGateAcrossToolsintests/test_safety_layer.py)Pending confirmations are held in memory and cleared on restart. (
_pending_confirmationsinresponse.py)EXECUTEDmeans Wazuh accepted the command and sent it to the agent. The API does not report whether the script succeeded; check/var/ossec/logs/active-responses.logon the agent.A command name such as
firewall-droponly works if the manager's<active-response>configuration enables it; otherwise Wazuh answers "The command used is not defined in the configuration." Prefix!to run the agent's script by name instead, for example!host-deny.wazuh_agent_commandruns an active-response script installed on the agent (sent as!<command>). Wazuh 4.x has no API for running an arbitrary shell command, whatever the tool description says. (run_active_responseinsrc/wazuh_mcp/client.py,tests/test_active_response_client.py)Whether a model stops at the confirmation prompt is measured by the eval in
evals/(deepseek-flash, three runs per wording, 6 October 2026 UTC).With the June wording, the model stopped for the user in 2 of 9 destructive cases. In the other 7 it called the tool again with
confirm=Trueand the token itself, saying the original request was approval or that the action was harmless.After the October 2026 rewording of the server prompt, the tool descriptions and the confirmation output, it stopped in 9 of 9.
The full text of each version and its scores are in
evals/PROMPT_HISTORY.md.
A prompt rule is not a control. The token is still visible to the model, which can still confirm on its own, and nine runs of one model say nothing about other models. Treat the MCP client's tool-approval prompt as the only human check.
Access control
RBAC is off unless
WAZUH_RBAC_ROLEis set.WAZUH_RBAC_POLICYon its own loads a policy but restricts nothing. (test_rbac_is_off_by_default,test_policy_without_role_allows_everything)The MCP endpoint has no built-in client authentication. (
FastMCP(...)insrc/wazuh_mcp/server.pyis created without an auth provider)
Docker and observability
docker-compose.ymlmounts./config/wazuh_manager/ossec.conf, which is not in the repository. The development and testing in October 2026 used Wazuh's officialwazuh-dockersingle-node deployment (v4.14.8) plus one agent container instead.docker-compose.ymlsetsWAZUH_MCP_HOST=127.0.0.1, which binds the loopback interface inside the container; the published port8000:8000cannot reach it. Not covered by a test.No tool calls the
record_*helpers insrc/wazuh_mcp/metrics.py, so the Prometheus counters listed under Observability stay at zero.src/wazuh_mcp/openapi.pyis not mounted by the server, so there is no/docsor/openapi.jsonroute. The checked-inopenapi.jsoncovers 22 of the 28 tools.
Evaluation
The model-behaviour eval in
evals/covers 15 cases and was added in October 2026.
π Production Hardening
This project ships with safe defaults, but the docker-compose demo stack disables several security features for ease of local testing. Do not use the demo config in production without these changes:
1. Enable TLS Everywhere
Set
WAZUH_INSECURE=falseβ the docker-compose now defaults tofalse.Re-enable the Wazuh Indexer security plugin β remove
DISABLE_SECURITY_PLUGIN=trueandplugins.security.disabled=truefrom thewazuh.indexerservice. Generate proper certificates instead.Set
FILEBEAT_SSL_VERIFICATION_MODE=fullon the Wazuh Manager.
2. Restrict the MCP Endpoint
The MCP server exposes an HTTP endpoint on port 8000. This endpoint has no built-in client authentication. In production:
Bind to
127.0.0.1if the AI client runs on the same host, ORPlace the MCP server behind a reverse proxy with mutual TLS / API key auth, OR
Use network-level controls (firewall rules, security groups) to restrict access.
3. Secure the Wazuh Indexer
The Indexer (port 9200) must be network-accessible from the MCP server.
Use TLS with certificate verification (
WAZUH_INSECURE=false).Store Indexer credentials in a secrets manager (Docker secrets, Kubernetes secrets, HashiCorp Vault) β never in plaintext
.envfiles in production.Consider IP whitelisting at the network/firewall layer.
4. Harden the Audit Log
The default audit log location is
~/.wazuh-mcp/audit.jsonl. In production, setWAZUH_AUDIT_LOG=/var/log/wazuh-mcp/audit.jsonl(or another persistent volume outside the home directory).For true immutability, ship audit logs to an external SIEM or use a write-once-read-many (WORM) filesystem.
5. RBAC: Enable It
RBAC is disabled by default (WAZUH_RBAC_ROLE is unset β all tools available).
In production, set WAZUH_RBAC_ROLE=analyst (or stricter) and configure tool
permissions in your AI client to match.
6. Version Compatibility
This server targets Wazuh 4.x (tested on 4.14.5). Wazuh 5.x replaces the Indexer with a new storage back-end β this server will require updates to work with 5.x. Check your Wazuh version before deploying.
π Security Policy
See docs/SECURITY.md for full defense-in-depth documentation (6 layers), production deployment checklist, and vulnerability reporting process.
π οΈ Troubleshooting
See docs/TROUBLESHOOTING.md for solutions to:
Wazuh dashboard version mismatch
/alertsreturning 404 (indexer setup)/rulesreturning 500 (Wazuh 4.14.x bug + indexer fallback)Indexer 401 authentication
Filebeat connection issues
MCP server connectivity
Rate limiting and confirmation gate behavior
Complete network architecture diagram
π License
MIT Β© Sbharadwaj05
Need custom detection rules, SIEM tuning, or an MCP server built? Topmate: https://topmate.io/subhash_bharadwaj or DM for Fiverr gig links.
Available Tools
28 toolswazuh_agent_commandA
β οΈ DESTRUCTIVE: Execute an arbitrary command on a remote Wazuh agent via the active-response infrastructure.
π SAFETY: Same confirmation flow as wazuh_run_active_response. You MUST confirm explicitly before the command runs.
Inputs: agent_id (required), command (required), confirm, confirmation_token. Returns: result
wazuh_agent_healthA
Get a fleet-wide health overview: counts by connection status, agents by OS/platform, version distribution, and stale agents. Use this for daily ops check or before an investigation.
Inputs: stale_threshold_hours. Returns: result
wazuh_alert_summaryA
Get a high-level summary of recent alerts: severity distribution, top attacking IPs, most triggered rules, and MITRE technique coverage. Use this as the first step in security posture assessment or shift handoff.
Inputs: hours_back, min_level. Returns: result
wazuh_cluster_node_statsA
Get detailed statistics for a specific Wazuh cluster node. Shows per-node EPS, queue sizes, daemon status, and resource utilization. Essential for diagnosing cluster imbalances.
Inputs: node_id (required). Returns: result
wazuh_cluster_statusA
Get the Wazuh cluster health status: node list, sync status, and connectivity between manager nodes. Use this when checking if the cluster is healthy or diagnosing replication failures.
Inputs: include_nodes. Returns: result
wazuh_compliance_reportA
Generate a compliance summary report across agents. Shows which agents have SCA enabled, their compliance scores, and failed-check counts grouped by policy. Ideal for audit prep.
Inputs: agent_ids. Returns: result
wazuh_get_agentA
Get detailed information about a specific agent: configuration, enabled modules, OS details, group membership, and connection history.
Inputs: agent_id (required). Returns: result
wazuh_get_alertA
Fetch a single Wazuh alert by its ID with full contextual detail. Use this when investigating a specific alert from wazuh_list_alerts results.
Inputs: alert_id (required). Returns: result
wazuh_get_cdb_listB
Read the contents of a specific CDB list. CDB lists are used for IP reputation, user whitelists, IOC matching, and more. Returns the key-value entries in the list.
Inputs: list_name (required), search, limit, offset. Returns: result
wazuh_get_groupA
Get detailed information about a specific agent group, including its configuration and member agents.
Inputs: group_id (required). Returns: result
wazuh_group_agentsB
List all agents belonging to a specific agent group.
Inputs: group_id (required), limit, offset. Returns: result
wazuh_incident_timelineA
After identifying a security incident, reconstruct a timeline of all related events leading to it. Takes an alert ID, traces back through related events on the same agent, and builds a chronological timeline of what happened.
Inputs: alert_id (required), lookback_hours, max_events. Returns: result
wazuh_list_agentsA
List all Wazuh agents with their status, OS, version, and last connection. Filter by status ('active', 'disconnected', 'never_connected'), search by name or IP, and control pagination.
Inputs: status, older_than, search, sort, +2 more. Returns: result
wazuh_list_alertsA
Query Wazuh security alerts with powerful filters. Use this to triage incidents, hunt for specific threat patterns, or get an overview of recent security events.
Inputs: agent_id, agents_list, min_level, rule_id, +6 more. Returns: result
wazuh_list_cdb_listsA
List all CDB (Constant Database) lists configured in Wazuh. CDB lists store key-value data used by rules β IP blocklists, user whitelists, IOC databases, etc. Essential for understanding what threat intelligence feeds are active.
Inputs: search, limit, offset. Returns: result
wazuh_list_groupsA
List all Wazuh agent groups. Groups are used to organize agents by function (e.g., 'web-servers', 'database', 'production'). Useful for scoping queries and active responses to specific agent sets.
Inputs: search, limit, offset. Returns: result
wazuh_manager_logsA
Retrieve Wazuh manager logs for troubleshooting. Filter by category ('ossec', 'api', 'all'), search for specific errors or warnings, and paginate through results.
Inputs: category, search, limit, offset. Returns: result
wazuh_manager_statsA
Retrieve Wazuh manager daemon statistics: events per second (EPS), queue sizes, processed events, and daemon health. Essential for capacity planning and troubleshooting performance issues.
Inputs: daemon. Returns: result
wazuh_query_fimA
Query File Integrity Monitoring (FIM) records. See what files were added, modified, or deleted on an agent. Essential for breach impact analysis and configuration drift detection.
Inputs: agent_id (required), file_path, event_type, search, +2 more. Returns: result
wazuh_query_vulnerabilitiesC
Query the Wazuh vulnerability-detector inventory. Find CVEs affecting your fleet, filtered by severity, agent, or specific CVE ID.
Inputs: agent_id (required), cve, severity, search, +2 more. Returns: result
wazuh_rules_coverage_mapA
Generate a coverage map showing which Wazuh rules map to which MITRE ATT&CK techniques, NIST 800-53 controls, PCI DSS requirements, GDPR articles, and HIPAA controls. Essential for compliance gap analysis and detection engineering.
Inputs: framework, min_level, limit. Returns: result
wazuh_rules_infoA
Search and list Wazuh detection rules. Filter by rule level, compliance framework (PCI DSS, GDPR, HIPAA, NIST 800-53), or MITRE ATT&CK technique. Essential for understanding your detection coverage and tuning rules.
Inputs: search, level, pci, gdpr, +5 more. Returns: result
wazuh_run_active_responseA
β οΈ DESTRUCTIVE: Trigger an active-response command on a Wazuh agent. Can block IPs via firewall, quarantine hosts, run custom scripts, etc.
π SAFETY: By default, this tool DOES NOT execute anything. It returns a confirmation prompt showing exactly what will happen. You MUST call it again with confirm=True and the correct confirmation_token to execute.
Inputs: agent_id (required), command (required), arguments, confirm, +1 more. Returns: result
wazuh_sca_checksA
Get detailed SCA check results β see exactly which compliance checks passed or failed on an agent. Filter by policy, search, or result status.
Inputs: agent_id (required), policy_id, result, search, +2 more. Returns: result
wazuh_sca_statusA
Get the Security Configuration Assessment (SCA) compliance status for an agent. Shows which policies are applied, pass/fail counts, and overall compliance scores.
Inputs: agent_id. Returns: result
wazuh_search_eventsA
Search raw security events across all Wazuh agents. Use this for deep threat hunting β search for IOCs like IPs, file hashes, commands, or process names in the raw event stream.
Inputs: search (required), select, sort, limit, +1 more. Returns: result
wazuh_search_mitreA
Search the MITRE ATT&CK framework as integrated with Wazuh. Look up techniques, find which Wazuh rules map to a technique, or discover what techniques are covered by your detection ruleset.
Inputs: search, technique_id, limit, offset. Returns: result
wazuh_vulnerability_heatmapA
Generate a vulnerability heatmap showing CVE severity distribution across agents. Identifies which systems have the most critical unpatched vulnerabilities. Essential for patch prioritization.
Inputs: severity. Returns: result
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelogβ¦
TDQS
Scored across 28 tools
Every tool has a clearly distinct purpose, from agent management to alert querying to compliance checks. Overlapping functions like listing vs. searching alerts are differentiated by scope and detail.
All tools follow a consistent 'wazuh_verb_noun' pattern (e.g., wazuh_list_agents, wazuh_get_alert), making it predictable and easy for an agent to infer action and resource.
With 28 tools, the set is comprehensive but still well-scoped for the Wazuh security management domain. While slightly above the typical 3-15 range, each tool addresses a specific function without redundancy.
The tool surface covers agent lifecycle, health, alerts, rules, compliance, vulnerability, FIM, cluster, logs, and active response. No obvious dead ends; destructive tools are paired with safety confirmation flows.
Maintenance
Related MCP Connectors
Official Microsoft MCP Server to query Microsoft Entra data using natural language
MCP server for Pentest-Tools.com: run scans, manage findings and reports via your preffered LLM.
Security scanner for MCP servers. Detect vulnerabilities, prompt injection, and tool poisoning.
AI-powered incident management and server monitoring via MCP.
Related MCP Servers
- AlicenseBqualityBmaintenanceAn MCP server for the Wazuh SIEM/XDR platform that enables users to query agents, security alerts, detection rules, and decoders through Claude or other MCP clients. It provides specialized tools and prompts for investigating security alerts, performing agent health checks, and generating environmental security overviews.2866 npm9MIT
- AlicenseNot gradedqualityBmaintenanceA production-ready Model Context Protocol (MCP) server for seamless integration between Wazuh SIEM and Large Language Models (LLMs).98AGPL 3.0
- AlicenseAqualityDmaintenanceMCP server that exposes OSSEC HIDS security monitoring capabilities as tools, resources, and prompts for AI assistants.262MIT
- AlicenseNot gradedqualityDmaintenanceAn MCP server that enables conversational interaction with Wazuh SIEM, allowing users to investigate alerts, hunt threats, tune false positives, edit rules, and run security actions via natural language.MIT