RAD Security MCP Server

A Model Context Protocol (MCP) server for RAD Security, providing AI-powered security insights for Kubernetes and cloud environments.

Connect (hosted — recommended)

RAD Security runs the MCP server for you, so most users don't need to install or host anything. Point your MCP client at the hosted endpoint and authenticate with your RAD Security credentials.

• Endpoint: https://api.rad.security/mcp/ — note the trailing slash.

• Transport: Streamable HTTP.

• Authentication: send your credential in the Authorization header:

  Code

  Copy

  Authorization: Bearer <access_key_id>:<secret_key>:<account_id>
  

  <access_key_id> and <secret_key> are a RAD Security API access key (create one in the RAD Security console); <account_id> is your account ID. The server authenticates every request against the RAD Security API — no credentials are stored server-side.

A short-lived form Bearer ory_st_<session_token>:<account_id> also works, but session tokens expire — prefer an access key for anything long-lived (e.g. Slack / Claude Tag).

Claude Code

claude mcp add --transport http rad-security https://api.rad.security/mcp/ \
  --header "Authorization: Bearer <access_key_id>:<secret_key>:<account_id>"

OpenAI Codex CLI

~/.codex/config.toml:

[mcp_servers.rad-security]
url = "https://api.rad.security/mcp/"
http_headers = { "Authorization" = "Bearer <access_key_id>:<secret_key>:<account_id>" }

Or via the CLI, keeping the secret in an env var (export RAD_MCP_TOKEN=<access_key_id>:<secret_key>:<account_id>):

codex mcp add rad-security --url https://api.rad.security/mcp/ --bearer-token-env-var RAD_MCP_TOKEN

Cursor

.cursor/mcp.json:

{
  "mcpServers": {
    "rad-security": {
      "type": "http",
      "url": "https://api.rad.security/mcp/",
      "headers": {
        "Authorization": "Bearer <access_key_id>:<secret_key>:<account_id>"
      }
    }
  }
}

VS Code (GitHub Copilot)

.vscode/mcp.json — note the wrapper key is servers, not mcpServers:

{
  "servers": {
    "rad-security": {
      "type": "http",
      "url": "https://api.rad.security/mcp/",
      "headers": {
        "Authorization": "Bearer <access_key_id>:<secret_key>:<account_id>"
      }
    }
  }
}

Gemini CLI

~/.gemini/settings.json — note the URL field is httpUrl (not url):

{
  "mcpServers": {
    "rad-security": {
      "httpUrl": "https://api.rad.security/mcp/",
      "headers": {
        "Authorization": "Bearer <access_key_id>:<secret_key>:<account_id>"
      }
    }
  }
}

Cline

cline_mcp_settings.json — note type must be exactly streamableHttp (camelCase):

{
  "mcpServers": {
    "rad-security": {
      "type": "streamableHttp",
      "url": "https://api.rad.security/mcp/",
      "headers": {
        "Authorization": "Bearer <access_key_id>:<secret_key>:<account_id>"
      }
    }
  }
}

Windsurf

~/.codeium/windsurf/mcp_config.json — note the URL field is serverUrl:

{
  "mcpServers": {
    "rad-security": {
      "serverUrl": "https://api.rad.security/mcp/",
      "headers": {
        "Authorization": "Bearer <access_key_id>:<secret_key>:<account_id>"
      }
    }
  }
}

Other clients

Most MCP clients accept a remote Streamable HTTP server with a URL and an Authorization header — only the field names differ. Keep the trailing slash on the URL in every case.

Claude.ai / Claude Desktop / Claude Tag (Slack)

These surfaces add remote MCP servers as connectors, which use their own credential settings rather than a raw request header. Add https://api.rad.security/mcp/ as a custom connector, then supply the bearer credential through the connector's settings:

• Claude Tag (Slack): attach the server as a plugin whose .mcp.json points at the endpoint, and add the bearer credential on the Access bundle's Credentials tab. See Claude Tag — connect a custom MCP server.
• Claude.ai / Desktop: add it under Settings → Connectors; see custom connectors.

Test it (MCP Inspector or curl)

npx @modelcontextprotocol/inspector
# Transport:      Streamable HTTP
# URL:            https://api.rad.security/mcp/   (trailing slash)
# Custom headers: { "Authorization": "Bearer <access_key_id>:<secret_key>:<account_id>" }

curl -H "authorization: Bearer <access_key_id>:<secret_key>:<account_id>" \
  -H "content-type: application/json" \
  -H "accept: application/json, text/event-stream" \
  -X POST https://api.rad.security/mcp/ \
  -d '{"jsonrpc":"2.0","id":1,"method":"initialize","params":{"protocolVersion":"2024-11-05","capabilities":{},"clientInfo":{"name":"curl","version":"1"}}}'

Scoping the tools an agent sees

By default a connection gets every toolkit. To give an agent a smaller set — less context/token overhead, and least privilege — add a scoping header to that connection alongside Authorization. The subset is enforced: an out-of-scope tool is hidden from tools/list and rejected if called.

Toolkits: containers, clusters, audit, images, kubeobject, runtime, findings, inbox, workflows, knowledge_base, radql, dashboards, integrations. All are enabled by default — narrow with the headers above, and use X-Rad-Readonly when you want to exclude every write tool.

Example — a read-only findings/images agent (any client that supports headers; Cursor shown):

{
  "mcpServers": {
    "rad-security-findings": {
      "type": "http",
      "url": "https://api.rad.security/mcp/",
      "headers": {
        "Authorization": "Bearer <access_key_id>:<secret_key>:<account_id>",
        "X-Rad-Toolkits": "findings, images",
        "X-Rad-Readonly": "true"
      }
    }
  }
}

In Claude Code, pass an extra --header:

claude mcp add --transport http rad-security https://api.rad.security/mcp/ \
  --header "Authorization: Bearer <access_key_id>:<secret_key>:<account_id>" \
  --header "X-Rad-Toolkits: findings, images"

Features

All tools require authentication and an account in RAD Security. The hosted endpoint exposes every toolkit below by default; scope a client down with X-Rad-Toolkits / X-Rad-Exclude-Toolkits, or drop all write tools with X-Rad-Readonly: true.

• Account Inventory

  • List clusters and their details

• Containers Inventory

  • List containers and their details

• Security Findings

  • List and analyze security findings
  • Update the status of a security finding

• Runtime Security

  • Get process trees of running containers
  • Get runtime baselines of running containers
  • Analyze process behavior of running containers

• Audit

  • List who shelled into a pod

• Images and Vulnerabilities

  • Get SBOMs
  • List images and their vulnerabilities
  • Get top vulnerable images
  • Ignore / unignore CVEs and list active CVE dispositions

• Kubernetes Objects

  • Get details of a specific Kubernetes resource
  • List Kubernetes resources

• Inbox

  • List inbox items and their details
  • Mark an inbox item as a false positive

• Automations (workflows)

  • List automations, runs and schedules
  • Get automation and run details
  • Run an automation
  • Create and update automations, and add cron schedules

  "Automation" is the product name users see; "workflow" is the underlying Windmill object the API and tool names use. They are the same thing.

• Knowledge Base

  • Search the knowledge base
  • List collections and documents
  • Run structured queries against a document

• Dashboards

  • List dashboards and get their details
  • List and get dashboard and widget templates
  • Create a dashboard, and update one in place (omitted fields are left unchanged, so a small edit does not require resending the whole dashboard)

• Integrations

  • List external integrations

• RadQL (Advanced Querying)

  • List available data types for querying (containers, findings, kubernetes_resources, etc.)
  • Get schema/metadata for specific data types
  • List possible values for filter fields
  • Execute RadQL queries with filtering, searching, and aggregations
  • Build queries programmatically from structured conditions
  • Execute multiple queries in parallel

Self-hosting

Prefer to run the server yourself — for example an air-gapped environment, data-residency requirements, or if you don't want to route through the hosted gateway? It's published to npm and as a container image.

Prerequisites