Flipper Feature Flags

Shipping a risky task? Wrap it in a Flipper flag and turn it on per user, per company, or for just your team — no redeploy required.

Two flavors matter for CMDx:

• Fail when off — callers get a failed Result (good for APIs that should return an error).
• Skip when off — callers get a skipped Result (good for “nice to have” background work).

Setup — fail when the flag is off

Middleware calls task.fail! before work. Runtime turns that into a failed result with your message and code.

class CmdxFlipperMiddleware
  def initialize(feature:, actor: :user)
    @feature = feature
    @actor   = actor
  end

  def call(task)
    actor = resolve_actor(task)

    unless Flipper.enabled?(@feature, actor)
      task.fail!("feature #{@feature} is disabled", code: :feature_disabled)
    end

    yield
  end

  private

  def resolve_actor(task)
    case @actor
    when nil    then nil
    when Symbol then task.context[@actor]
    when Proc   then task.instance_exec(&@actor)
    else             @actor.respond_to?(:call) ? @actor.call(task) : @actor
    end
  end
end

Usage — fail when off

class RebuildSearchIndex < CMDx::Task
  register :middleware, CmdxFlipperMiddleware.new(feature: :search_v2, actor: :company)

  required :company

  def work
    SearchIndex.rebuild!(company)
  end
end

Usage — skip when off

When “off” means “do nothing, but that's OK,” halt inside work with skip! so the result reports skipped?:

class RebuildSearchIndex < CMDx::Task
  required :company

  def work
    skip!("search_v2 disabled for #{company.id}") unless Flipper.enabled?(:search_v2, company)

    SearchIndex.rebuild!(company)
  end
end

Notes

Skip from middleware too

Middleware can throw success!, skip!, fail!, or throw! — Runtime catches them via catch(Signal::TAG). Use task.skip!(...) before yield when you want a skipped result without entering work. Throw before yielding; signals thrown after the lifecycle finishes are ignored.