SWAG Skill

Mode Detection

MCP mode (preferred): Use when mcp__swag-mcp__swag tool is available. The server manages nginx proxy configuration files directly (local or remote via SSH).

HTTP fallback: No meaningful curl equivalent — SWAG config management requires direct filesystem access. If MCP server is unavailable, surface the issue to the user and suggest restarting it.

MCP URL: ${user_config.swag_mcp_url}

MCP Mode — Tool Reference

Single tool: mcp__swag-mcp__swag with an action parameter.

List configurations

mcp__swag-mcp__swag
  action: "list"

Returns all proxy configurations with their status (enabled/disabled).

Create configuration

mcp__swag-mcp__swag
  action:       "create"
  server_name:  (required) Service name, e.g. "jellyfin", "sonarr"
  type:         (required) "subdomain" or "subfolder"
  upstream_url: (required) Backend URL, e.g. "http://192.168.1.10:8096"
  auth_method:  (optional) "authelia", "authentik", "basic", "none" — default from server config
  enable_quic:  (optional) true/false

View configuration

mcp__swag-mcp__swag
  action:      "view"
  server_name: (required) Service name

Edit configuration

mcp__swag-mcp__swag
  action:      "edit"
  server_name: (required) Service name
  changes:     (required) Description of changes to apply

Update upstream

mcp__swag-mcp__swag
  action:       "update"
  server_name:  (required) Service name
  upstream_url: (required) New backend URL

Remove configuration

mcp__swag-mcp__swag
  action:      "remove"
  server_name: (required) Service name

DESTRUCTIVE — removes the nginx config file. Always confirm with user before executing.

View logs

mcp__swag-mcp__swag
  action:   "logs"
  log_type: (optional) "access", "error", "fail2ban", "letsencrypt" — default "error"

Manage backups

mcp__swag-mcp__swag
  action: "backups"

Lists available configuration backups.

Health check

mcp__swag-mcp__swag
  action:      "health_check"
  server_name: (optional) Check specific service — omit for all

Probes whether proxied services are accessible.

Typical Workflows

Expose a new service

1. action: "list" — confirm no existing config for the service
2. action: "create" — create the config with upstream URL and auth method
3. action: "health_check" — verify the service is accessible

Update a service's backend

1. action: "view" — confirm current config
2. action: "update" — set new upstream URL
3. action: "health_check" — verify

Diagnose issues

1. action: "health_check" — check what's failing
2. action: "logs" — review error/access logs

Destructive Operations

Always confirm before:

• action: "remove" — permanently deletes the proxy config
• action: "edit" — modifies existing config (non-reversible without backup)

Proxy Confs Path

The server manages configs at the path configured via SWAG_MCP_PROXY_CONFS_PATH or the SSH URI in SWAG_MCP_PROXY_CONFS_URI. Plugin userConfig is projected through swag setup repair into ~/.swag-mcp/.env; existing local secrets are preserved unless a replacement option is supplied.

Notes

• subdomain configs require a wildcard DNS entry or per-subdomain record pointing to SWAG
• subfolder configs proxy via path prefix (e.g. https://domain.com/sonarr/)
• Auth methods: authelia and authentik require those services to be running and configured
• QUIC/HTTP3 requires ports 443/UDP to be open in addition to 443/TCP