Connect Wazuh to Gemini CLI via MCP: Agents and Files

💡 No install? Use cloud.anythingmcp.com directly. Sign in, click Connectors → Wazuh, paste your credentials, mint an MCP API key — done. No Docker, no git clone, no local server.

Wazuh + Gemini

Let your AI read your Wazuh 4.x security platform through the server API: manager version and daemon status, agents and their connection status, the detection rules and decoders, Security Configuration Assessment (CIS benchmark) results and File Integrity Monitoring data per agent. Read-only: it never restarts, changes or deletes anything. Alerts and vulnerabilities live in the Wazuh indexer and are not covered.

Prerequisites

See the full setup instructions baked into the connector (visible in the in-app store when you select the connector). The required environment variables for this connector are:

WAZUH_URL, WAZUH_USERNAME, WAZUH_PASSWORD

Step 1 — Get credentials

Setting up takes about 5 minutes. You need a Wazuh 4.x server API (the manager, port 55000 by default) that AnythingMCP can reach with a certificate it trusts, and an API user with the built-in readonly role.

1. Set WAZUH_URL to the server API address, normally https://<manager-host>:55000, without a trailing slash. Not the dashboard (port 443) and not the indexer (port 9200).
2. Create a dedicated API user for the AI and give it the built-in readonly role: in the Wazuh dashboard under Server management → Security → Users, or with POST /security/users and POST /security/users/{user_id}/roles.
3. Put that user's name in WAZUH_USERNAME and its password in WAZUH_PASSWORD. The connector signs in with them and renews the session token itself (Wazuh tokens last 900 seconds by default).
4. Install the connector and run wazuh_get_manager_info: it returns the manager version and confirms that address and sign-in work.

Read-only: every tool is a GET request. The connector cannot restart anything, run active responses, add, remove or upgrade agents, or edit rules, decoders, users or configuration. The readonly role covers every permission it needs (manager:read, agent:read, rules:read, decoders:read, sca:read and syscheck:read).

What it reads: manager and API versions and the state of each manager daemon; agents with status, IP, OS, version, groups and last keep-alive, counts by status (wazuh_get_agents_summary), OS platforms and agents older than the manager; the ruleset by level, group, compliance mapping (PCI DSS, GDPR, HIPAA, NIST 800-53, TSC) or MITRE id, and the decoders; Security Configuration Assessment (CIS benchmark) policies and failed checks per agent, with rationale and remediation; File Integrity Monitoring (syscheck) files, hashes and the last scan per agent. Agent ids have at least three digits (001); 000 is the manager itself.

Certificate: Wazuh installs a self-signed certificate on port 55000, and a self-signed certificate fails: the connector has no switch to trust any certificate. Give the API a certificate whose name matches WAZUH_URL (https.cert and https.key in Wazuh's api.yaml), put a reverse proxy with a valid certificate in front of it, or, on self-hosted, start AnythingMCP with NODE_EXTRA_CA_CERTS pointing to your internal CA.

Where Wazuh runs: on a self-hosted AnythingMCP, a Wazuh on a private address (https://10.0.0.5:55000, https://wazuh.intern:55000) needs an administrator to allow the host in SSRF_ALLOWED_HOSTS or in the allowlist in the admin settings. AnythingMCP Cloud can only reach a Wazuh API exposed on the public internet over HTTPS with a publicly trusted certificate.

Privacy: agent names, IP addresses, operating systems, file paths and hashes describe your infrastructure. Assign the connector only to MCP servers whose users may see that.

Limits: list tools return 50 items by default and at most 500 per call. Wazuh allows 300 requests per minute by default and answers 429 above that, and by default 50 failed logins within 5 minutes block the address for 300 seconds.

Common errors: 401 "Invalid credentials" means a wrong WAZUH_USERNAME or WAZUH_PASSWORD. 403 means the API user lacks a read permission: give it the readonly role. 404 on every call means WAZUH_URL points at the dashboard or the indexer instead of the server API. "Agent does not exist" (Wazuh error 1701) means no agent has that id. 400 "There is no database for agent" means the agent never connected, so it has no SCA or FIM data yet. "self-signed certificate" or "unable to verify the first certificate" is the certificate problem above.

The connector does not read alerts and events or vulnerability detection results (they live in the Wazuh indexer, not in this API), does not write anything, does not cover cluster configuration, and does not support Wazuh 5.x, which removes most of these endpoints.

Step 2 — Install the adapter

curl -fsSL https://raw.githubusercontent.com/HelpCode-ai/anythingmcp/main/docker-compose.quickstart.yml -o docker-compose.yml
printf 'JWT_SECRET=%s\nENCRYPTION_KEY=%s\n' "$(openssl rand -hex 32)" "$(openssl rand -hex 32)" > .env
docker compose up -d

Step 3 — Add the connector in Gemini

Your server URL: in AnythingMCP, open MCP Servers → the server this connector is on, and copy its URL (https://cloud.anythingmcp.com/mcp/…). Use it wherever this guide shows YOUR_SERVER_ID.

Gemini CLI reads MCP servers from ~/.gemini/settings.json (on Windows %USERPROFILE%\.gemini\settings.json). Add:

{
  "mcpServers": {
    "anythingmcp": {
      "httpUrl": "https://cloud.anythingmcp.com/mcp/YOUR_SERVER_ID",
      "headers": { "X-API-Key": "YOUR_MCP_API_KEY" }
    }
  }
}

1. Create an MCP API key in AnythingMCP under MCP Servers → your server → API keys, and paste it in place of YOUR_MCP_API_KEY.
2. Save the file and restart gemini.
3. Run /mcp inside Gemini CLI (or gemini mcp list in your shell) — anythingmcp should show as connected.
4. Start prompting — Gemini CLI can now call the Wazuh tools.

Available tools

FAQ

Where does Gemini CLI look for MCP servers? In ~/.gemini/settings.json for your user, or in .gemini/settings.json inside a project. gemini mcp list shows each server and whether it connects.

Next steps

• Connector landing page
• Connect to Claude
• Connect to ChatGPT