Most tools target distinct resource-action pairs, so an agent can usually select correctly. A few compose and container tools could be confused (compose_ps vs list_containers, compose_logs vs container_logs), but their descriptions and compose prefix make the boundary clear.

Many tools follow verb_noun (list_containers, build_image, prune_images), but several use noun_noun (container_logs, system_info, security_scan) or noun_verb (network_connect). The mix is readable but not a consistent pattern.

27 tools is past the typical MCP range and creates a heavy selection surface. Docker is broad, but the compose and container categories overlap enough that the count feels too high.

Core operations exist for containers, images, networks, and volumes, but common lifecycle operations are missing: a stopped container cannot be started, a specific image cannot be removed, and there is no network disconnect. These gaps will cause agent failures on everyday Docker tasks.