kubeswarm MCP Server Integration for Agents

kubeswarm agents connect to external tool servers via the Model Context Protocol (MCP) SSE transport. Configure auth, headers and health monitoring declaratively in YAML.

Configuration

spec:
  tools:
    mcp:
      - name: github
        url: https://mcp-github.example.com/sse
        trust: internal
        instructions: "Read-only. Branch: main."
        auth:
          bearer:
            secretKeyRef:
              name: mcp-tokens
              key: github-token
        headers:
          - name: X-Project-ID
            value: "my-project"

Authentication

Method	Configuration	Use case
None	Omit auth	Internal/development servers
Bearer	auth.bearer.secretKeyRef	Token-based auth
mTLS	auth.mtls.secretRef	Certificate-based auth

Bearer and mTLS are mutually exclusive. The namespace security policy can require auth on all MCP servers - see MCP Policy.

Health Monitoring

The operator probes each MCP server with an HTTP GET every 60 seconds:

• Any response with status < 500 = healthy (including 401 - reachable but needs auth)
• 5xx or connection error = unhealthy, sets MCPDegraded condition

kubectl describe swagent my-agent | grep -A5 "MCP Servers"

Multiple servers

Agents can connect to multiple MCP servers. Each is independent:

spec:
  tools:
    mcp:
      - name: github
        url: https://mcp-github.example.com/sse
        trust: internal
      - name: jira
        url: https://mcp-jira.example.com/sse
        trust: internal
      - name: search
        url: https://mcp-search.example.com/sse
        trust: external

Tool names are prefixed with the server name: github/read_file, jira/create_issue.