completion-kit 0.18.1 → 0.20.0

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 620084f7c112f139684433d2d6f2d8ee407bfd5c7aaf1612ac4a5334c404e490
-  data.tar.gz: 7e93349d581b10eaa993250d678f80550492a0075c426b8970a97c498b277ce6
+  metadata.gz: 44b946858390ad2aa5d3a3bcb2a68ba666a6725ba5e77725887f99087603e322
+  data.tar.gz: 52499c569d0a73b0b714547e9aeed1535abe452cfd093288356d1580ee83d7ce
 SHA512:
-  metadata.gz: e94ead519660a768f88f7981128f338f39448837ad12cf45ca26409a7b574bc9d6a2e6980f2fef839c5fbd44d17ad465b4cce21937068a98db8312e06f437dde
-  data.tar.gz: 4f31db3df58c983066fbd19b71f8f978215755f592f8637e0116a1b3f0fb33c213079517503077cda2622c52895531c0226f0407f0c32ba33d0788c134bae64f
+  metadata.gz: 406ae145ee5963e7a1a1a32b7669591f8c04ae49cb1166d6e054d657a77aa0fca218c78a4d2f7b31dc5518ee9aa75a7f5bae28889f81753ccdcdf86822b4dfb7
+  data.tar.gz: 2c9f542f56209a7fde4d8d6cc822e7096b43b178d7fef9ef5c65cc95e971c512ad91b14c1713fefd7d076164bff99e74ec587600c9900ce55994e67ff4a273b2

data/Rakefile

@@ -7,3 +7,10 @@ RSpec::Core::RakeTask.new(:spec) do |t|
 end
 
 task default: :spec
+
+desc "Run the full suite with judge API keys cleared, matching CI's keyless environment"
+task :release_guard do
+  sh "OPENAI_API_KEY= ANTHROPIC_API_KEY= OLLAMA_API_KEY= bundle exec rspec"
+end
+
+Rake::Task["release:guard_clean"].enhance([:release_guard])

data/app/assets/stylesheets/completion_kit/application.css

@@ -413,6 +413,10 @@ form.button_to {
   grid-template-columns: repeat(3, minmax(0, 1fr));
 }
 
+.ck-grid--cards-4 {
+  grid-template-columns: repeat(4, minmax(0, 1fr));
+}
+
 /* ── Dashboard: workspace stat ribbon ─────────────────────────────────
    Replaces the old oversized count cards. One thin instrument strip,
    four navigable segments split by hairline dividers. */
@@ -3698,7 +3702,7 @@ table.ck-runs-table {
 }
 
 .ck-runs-table td {
-  vertical-align: middle;
+  vertical-align: top;
   padding-top: 0.7rem;
   padding-bottom: 0.7rem;
 }
@@ -5163,6 +5167,7 @@ a.tag-mark {
 .ck-failure-list__surface--run { color: var(--ck-warning); }
 .ck-failure-list__surface--generation { color: var(--ck-danger); }
 .ck-failure-list__surface--judge { color: var(--ck-info); }
+.ck-failure-list__surface--check { color: var(--ck-danger); }
 .ck-failure-list__cause {
   overflow: hidden;
   text-overflow: ellipsis;

data/app/controllers/completion_kit/api/v1/imports_controller.rb

@@ -0,0 +1,29 @@
+module CompletionKit
+  module Api
+    module V1
+      class ImportsController < BaseController
+        def promptfoo
+          content = params[:config].presence || request.raw_post
+          result = PromptfooImporter.call(content)
+
+          if result.ok
+            render json: import_summary(result), status: :created
+          else
+            render_error(result.error, status: :unprocessable_entity)
+          end
+        end
+
+        private
+
+        def import_summary(result)
+          {
+            prompts: result.prompts,
+            dataset: result.dataset,
+            metrics: result.metrics,
+            providers: result.providers
+          }
+        end
+      end
+    end
+  end
+end

data/app/controllers/completion_kit/api/v1/runs_controller.rb

@@ -134,11 +134,15 @@ module CompletionKit
                 {
                   metric_id: mid,
                   metric_name: anchor.metric_name,
+                  kind: anchor.check? ? "check" : "llm_judge",
                   left_score: l_review ? l_review.ai_score : nil,
                   right_score: r_review ? r_review.ai_score : nil,
+                  left_passed: l_review&.passed,
+                  right_passed: r_review&.passed,
                   left_metric_version_id: l_review&.metric_version_id,
                   right_metric_version_id: r_review&.metric_version_id,
-                  delta: (l_review&.ai_score && r_review&.ai_score) ? (r_review.ai_score.to_f - l_review.ai_score.to_f).round(2) : nil
+                  delta: (l_review&.ai_score && r_review&.ai_score) ? (r_review.ai_score.to_f - l_review.ai_score.to_f).round(2) : nil,
+                  result_change: CompletionKit::RunComparison.result_change(l_review&.passed, r_review&.passed)
                 }
               end.compact
             }

data/app/controllers/completion_kit/dashboard_controller.rb

@@ -14,6 +14,7 @@ module CompletionKit
       @activity = DashboardStats.activity
       @worst_metric = DashboardStats.worst_metric(since: 7.days.ago)
       @failures = DashboardStats.failures(since: 7.days.ago)
+      @failing_checks = DashboardStats.failing_checks(since: 7.days.ago)
       @ignored_metrics = DashboardDismissal.metrics
       @ignored_failures = DashboardDismissal.failures
       @prompt_changes = DashboardStats.prompt_changes

data/app/controllers/completion_kit/dashboard_dismissals_controller.rb

@@ -27,6 +27,7 @@ module CompletionKit
 
     def baseline_for(record)
       return nil unless record.is_a?(Metric)
+      return DashboardStats.metric_pass_rate(record.id, since: WINDOW.ago) if record.check?
       DashboardStats.metric_average(record.id, since: WINDOW.ago)
     end
 

data/app/controllers/completion_kit/imports_controller.rb

@@ -0,0 +1,31 @@
+module CompletionKit
+  class ImportsController < ApplicationController
+    def new
+    end
+
+    def create
+      content = uploaded_content
+
+      if content.blank?
+        flash.now[:alert] = "Paste or upload a promptfooconfig.yaml to import."
+        return render :new, status: :unprocessable_entity
+      end
+
+      @result = PromptfooImporter.call(content)
+
+      if @result.ok
+        render :create
+      else
+        flash.now[:alert] = @result.error
+        render :new, status: :unprocessable_entity
+      end
+    end
+
+    private
+
+    def uploaded_content
+      file = params[:file]
+      file.respond_to?(:read) ? file.read : params[:config]
+    end
+  end
+end

data/app/controllers/completion_kit/responses_controller.rb

@@ -1,5 +1,6 @@
 module CompletionKit
   class ResponsesController < ApplicationController
+    include CompletionKit::ResponseOrdering
     before_action :set_run
     before_action :set_response
 
@@ -24,21 +25,7 @@ module CompletionKit
     end
 
     def ordered_response_ids
-      if @run.judge_configured? && @sort == "score_asc"
-        @run.responses
-          .left_joins(:reviews)
-          .group("completion_kit_responses.id")
-          .order(Arel.sql("AVG(completion_kit_reviews.ai_score) ASC NULLS LAST"))
-          .pluck(:id)
-      elsif @run.judge_configured? && @sort != "none"
-        @run.responses
-          .left_joins(:reviews)
-          .group("completion_kit_responses.id")
-          .order(Arel.sql("AVG(completion_kit_reviews.ai_score) DESC NULLS LAST"))
-          .pluck(:id)
-      else
-        @run.responses.order(:id).pluck(:id)
-      end
+      ordered_responses_relation(@run, @sort).pluck(:id)
     end
   end
 end

data/app/controllers/completion_kit/runs_controller.rb

@@ -1,6 +1,7 @@
 module CompletionKit
   class RunsController < ApplicationController
     include CompletionKit::TagFiltering
+    include CompletionKit::ResponseOrdering
     before_action :set_run, only: [:show, :edit, :update, :destroy, :generate, :suggest, :retry_failures, :rerun, :regrade, :refresh_status, :compare]
     before_action :load_form_collections, only: [:new, :edit, :create, :update]
 
@@ -10,21 +11,7 @@ module CompletionKit
     end
 
     def show
-      @responses = if @run.judge_configured? && params[:sort] == "score_asc"
-                     @run.responses
-                       .left_joins(:reviews)
-                       .includes(:reviews)
-                       .group("completion_kit_responses.id")
-                       .order(Arel.sql("AVG(completion_kit_reviews.ai_score) ASC NULLS LAST"))
-                   elsif @run.judge_configured?
-                     @run.responses
-                       .left_joins(:reviews)
-                       .includes(:reviews)
-                       .group("completion_kit_responses.id")
-                       .order(Arel.sql("AVG(completion_kit_reviews.ai_score) DESC NULLS LAST"))
-                   else
-                     @run.responses.includes(:reviews).order(:id)
-                   end
+      @responses = ordered_responses_relation(@run, params[:sort]).includes(:reviews)
     end
 
     def new
@@ -208,11 +195,15 @@ module CompletionKit
             {
               metric_id: mid,
               metric_name: anchor.metric_name,
+              kind: anchor.check? ? "check" : "llm_judge",
               left_score: l_review ? l_review.ai_score : nil,
               right_score: r_review ? r_review.ai_score : nil,
+              left_passed: l_review&.passed,
+              right_passed: r_review&.passed,
               left_version_label: version_label_for(l_review, metric_versions),
               right_version_label: version_label_for(r_review, metric_versions),
-              delta: (l_review&.ai_score && r_review&.ai_score) ? (r_review.ai_score.to_f - l_review.ai_score.to_f).round(2) : nil
+              delta: (l_review&.ai_score && r_review&.ai_score) ? (r_review.ai_score.to_f - l_review.ai_score.to_f).round(2) : nil,
+              result_change: RunComparison.result_change(l_review&.passed, r_review&.passed)
             }
           end.compact
         }

data/app/controllers/concerns/completion_kit/response_ordering.rb

@@ -0,0 +1,25 @@
+module CompletionKit
+  module ResponseOrdering
+    extend ActiveSupport::Concern
+
+    private
+
+    FAILED_CHECKS_SQL = "SUM(CASE WHEN completion_kit_reviews.passed IS FALSE THEN 1 ELSE 0 END)".freeze
+    RUBRIC_AVG_SQL = "AVG(completion_kit_reviews.ai_score)".freeze
+
+    def ordered_responses_relation(run, sort)
+      return run.responses.order(:id) unless run.gradable?
+
+      composite = if sort == "score_asc"
+                    "#{FAILED_CHECKS_SQL} DESC, #{RUBRIC_AVG_SQL} ASC NULLS LAST"
+                  else
+                    "#{FAILED_CHECKS_SQL} ASC, #{RUBRIC_AVG_SQL} DESC NULLS LAST"
+                  end
+
+      run.responses
+         .left_joins(:reviews)
+         .group("completion_kit_responses.id")
+         .order(Arel.sql("#{composite}, completion_kit_responses.id ASC"))
+    end
+  end
+end

data/app/helpers/completion_kit/application_helper.rb

@@ -159,6 +159,34 @@ module CompletionKit
       :low
     end
 
+    def ck_check_badge(passed)
+      if passed == true
+        content_tag(:span, "Pass", class: ck_badge_classes(:high))
+      elsif passed == false
+        content_tag(:span, "Fail", class: ck_badge_classes(:low))
+      else
+        content_tag(:span, "Pending", class: ck_badge_classes(:pending))
+      end
+    end
+
+    def ck_result_change_badge(change)
+      case change
+      when "broke"
+        content_tag(:span, "Broke", class: "ck-delta ck-delta--negative")
+      when "fixed"
+        content_tag(:span, "Fixed", class: "ck-delta ck-delta--positive")
+      when "same"
+        content_tag(:span, "Same", class: "ck-delta ck-delta--zero")
+      end
+    end
+
+    def ck_pass_rate_kind(rate)
+      return :high if rate >= 0.9
+      return :medium if rate >= 0.7
+
+      :low
+    end
+
     def ck_word_diff_old(old_text, new_text)
       diff_tokens(old_text, new_text, :old)
     end

data/app/models/completion_kit/response.rb

@@ -63,15 +63,21 @@ module CompletionKit
     end
 
     def broadcast_row_update
-      run.broadcast_response_update(self)
+      safely_broadcast { run.broadcast_response_update(self) }
     end
 
     def broadcast_run_progress
-      run.broadcast_progress
+      safely_broadcast { run.broadcast_progress }
     end
 
     def should_broadcast_progress?
       saved_change_to_status? && terminal?
     end
+
+    def safely_broadcast
+      yield
+    rescue StandardError => e
+      Rails.logger.error("[CompletionKit] response ##{id} broadcast failed: #{e.class}: #{e.message}")
+    end
   end
 end

data/app/models/completion_kit/review.rb

@@ -40,15 +40,21 @@ module CompletionKit
     private
 
     def broadcast_parent_row_update
-      response.run.broadcast_response_update(response)
+      safely_broadcast { response.run.broadcast_response_update(response) }
     end
 
     def broadcast_run_progress
-      response.run.broadcast_progress
+      safely_broadcast { response.run.broadcast_progress }
     end
 
     def should_broadcast_progress?
       saved_change_to_status? && terminal?
     end
+
+    def safely_broadcast
+      yield
+    rescue StandardError => e
+      Rails.logger.error("[CompletionKit] review ##{id} broadcast failed: #{e.class}: #{e.message}")
+    end
   end
 end

data/app/services/completion_kit/dashboard_stats.rb

@@ -58,6 +58,28 @@ module CompletionKit
       scored_reviews_since(since).where(metric_id: metric_id).average(:ai_score)&.to_f&.round(2)
     end
 
+    def self.metric_pass_rate(metric_id, since:)
+      resolved = Review.joins(:response)
+                       .where(metric_id: metric_id)
+                       .where("completion_kit_reviews.created_at >= ?", since)
+                       .where.not(passed: nil)
+      total = resolved.count
+      return nil if total.zero?
+
+      (resolved.where(passed: true).count.to_f / total).round(2)
+    end
+
+    def self.failing_checks(since:)
+      reviews = Review.where(passed: false)
+                      .where("completion_kit_reviews.created_at >= ?", since)
+                      .includes(response: :run)
+                      .order(updated_at: :desc)
+      items = reviews.map do |review|
+        { metric_name: review.metric_name, response: review.response, run: review.response.run }
+      end
+      { count: items.size, items: items }
+    end
+
     # Everything that terminally failed in the window across all three
     # surfaces — failed runs, failed generations, failed judge reviews —
     # excluding any the user has dismissed. Returns a count and an items list

data/app/services/completion_kit/mcp_dispatcher.rb

@@ -36,7 +36,8 @@ module CompletionKit
         McpTools::ProviderCredentials.definitions +
         McpTools::Tags.definitions +
         McpTools::Agreements.definitions +
-        McpTools::Judges.definitions
+        McpTools::Judges.definitions +
+        McpTools::Imports.definitions
     end
 
     def self.call_tool(name, arguments)
@@ -52,6 +53,7 @@ module CompletionKit
       when /\Atags_/                 then McpTools::Tags.call(name, arguments)
       when /\Aagreements_/           then McpTools::Agreements.call(name, arguments)
       when /\Ajudges_/               then McpTools::Judges.call(name, arguments)
+      when /\Apromptfoo_/            then McpTools::Imports.call(name, arguments)
       else raise MethodNotFound, "Unknown tool: #{name}"
       end
     end

data/app/services/completion_kit/mcp_tools/imports.rb

@@ -0,0 +1,33 @@
+module CompletionKit
+  module McpTools
+    module Imports
+      extend Base
+
+      TOOLS = {
+        "promptfoo_import" => {
+          description: "Import a promptfooconfig.yaml. Creates a prompt, a dataset from the test vars, and metrics from the assert blocks (llm-rubric/g-eval become judge metrics; contains/equals/regex/is-json become deterministic check metrics). Returns a summary of what mapped and what was skipped and why; nothing is dropped silently.",
+          inputSchema: {
+            type: "object",
+            properties: {
+              config: {type: "string", description: "The full promptfooconfig.yaml contents."}
+            },
+            required: ["config"]
+          },
+          handler: :promptfoo_import
+        }
+      }.freeze
+
+      def self.promptfoo_import(args)
+        result = PromptfooImporter.call(args["config"])
+        return error_result(result.error) unless result.ok
+
+        text_result(
+          prompts: result.prompts,
+          dataset: result.dataset,
+          metrics: result.metrics,
+          providers: result.providers
+        )
+      end
+    end
+  end
+end

data/app/services/completion_kit/promptfoo_importer.rb

@@ -0,0 +1,174 @@
+require "yaml"
+require "csv"
+
+module CompletionKit
+  class PromptfooImporter
+    Result = Data.define(:ok, :error, :prompts, :dataset, :metrics, :providers)
+
+    JUDGE_ASSERTS = %w[llm-rubric g-eval model-graded-closedqa factuality].freeze
+
+    def self.call(content)
+      new(content).call
+    end
+
+    def initialize(content)
+      @content = content.to_s
+    end
+
+    def call
+      config = parse
+      return failure("Could not parse YAML: #{@parse_error}") if config.nil?
+      return failure("Top-level YAML must be a mapping of promptfoo config keys.") unless config.is_a?(Hash)
+
+      ApplicationRecord.transaction do
+        providers = import_providers(config)
+        prompts = import_prompts(config, default_model(config))
+        dataset = import_dataset(config)
+        metrics = import_metrics(config)
+        Result.new(ok: true, error: nil, prompts: prompts, dataset: dataset, metrics: metrics, providers: providers)
+      end
+    end
+
+    private
+
+    def parse
+      YAML.safe_load(@content, aliases: true)
+    rescue Psych::Exception => e
+      @parse_error = e.message
+      nil
+    end
+
+    def failure(message)
+      Result.new(ok: false, error: message, prompts: nil, dataset: nil, metrics: nil, providers: nil)
+    end
+
+    def import_providers(config)
+      entries = Array(config["providers"]).map { |p| provider_name(p) }.compact
+      configured = ProviderCredential.pluck(:provider).to_set
+      matched = []
+      unmatched = []
+      entries.each do |entry|
+        provider = entry.split(":").first
+        (configured.include?(provider) ? matched : unmatched) << entry
+      end
+      { matched: matched.uniq, unmatched: unmatched.uniq }
+    end
+
+    def provider_name(provider)
+      provider.is_a?(Hash) ? provider["id"] : provider
+    end
+
+    def default_model(config)
+      first = Array(config["providers"]).map { |p| provider_name(p) }.compact.first
+      return "gpt-4o" if first.nil?
+
+      parts = first.split(":")
+      parts.length > 1 ? parts.last : first
+    end
+
+    def import_prompts(config, model)
+      created = []
+      skipped = []
+      Array(config["prompts"]).each_with_index do |raw, index|
+        if raw.is_a?(String) && raw.start_with?("file://")
+          skipped << { value: raw, reason: "prompt file reference; paste the file's contents to import it" }
+          next
+        end
+        template = prompt_template(raw)
+        if template.nil?
+          skipped << { value: raw.inspect, reason: "unsupported prompt shape (only inline string templates import)" }
+          next
+        end
+
+        prompt = Prompt.create!(name: "Imported prompt #{index + 1}", template: template, llm_model: model)
+        created << prompt.name
+      end
+      { created: created, skipped: skipped }
+    end
+
+    def prompt_template(raw)
+      return raw if raw.is_a?(String)
+      raw["raw"] || raw["template"] if raw.is_a?(Hash)
+    end
+
+    def import_dataset(config)
+      tests = Array(config["tests"])
+      vars_rows = tests.map { |t| (t.is_a?(Hash) ? t["vars"] : nil) || {} }.select { |v| v.is_a?(Hash) }
+      return { skipped: "no tests with vars to import" } if vars_rows.empty?
+
+      columns = vars_rows.flat_map(&:keys).uniq
+      csv = CSV.generate do |out|
+        out << columns
+        vars_rows.each { |vars| out << columns.map { |c| vars[c] } }
+      end
+      dataset = Dataset.create!(name: "Imported dataset", csv_data: csv)
+      { created: dataset.name, rows: vars_rows.length, columns: columns }
+    end
+
+    def import_metrics(config)
+      created = []
+      skipped = []
+      asserts(config).each do |assert|
+        attrs = metric_attributes(assert)
+        if attrs.nil?
+          skipped << { type: assert["type"], reason: "no CompletionKit metric maps to this assert type" }
+          next
+        end
+
+        metric = Metric.create!(attrs)
+        created << { name: metric.name, type: metric.metric_type }
+      rescue ActiveRecord::RecordInvalid => e
+        skipped << { type: assert["type"], reason: e.record.errors.full_messages.join(", ") }
+      end
+      { created: created, skipped: skipped }
+    end
+
+    def asserts(config)
+      default = config.dig("defaultTest", "assert")
+      per_test = Array(config["tests"]).flat_map { |t| t.is_a?(Hash) ? Array(t["assert"]) : [] }
+      (Array(default) + per_test).select { |a| a.is_a?(Hash) && a["type"] }.uniq
+    end
+
+    def metric_attributes(assert)
+      type = assert["type"].to_s
+      value = assert["value"]
+      return { name: unique_name("Rubric"), metric_type: "llm_judge", instruction: value.to_s } if JUDGE_ASSERTS.include?(type)
+
+      config, label = check_mapping(type, value.to_s)
+      return nil if config.nil?
+
+      { name: unique_name(label), metric_type: "check", check_config: config }
+    end
+
+    def check_mapping(type, text)
+      case type
+      when "contains"
+        [{ "check_kind" => "contains", "target" => "response_text", "value" => text }, "Contains #{text.truncate(30).inspect}"]
+      when "icontains"
+        [{ "check_kind" => "contains", "target" => "response_text", "value" => text, "case_sensitive" => false }, "Contains #{text.truncate(30).inspect}"]
+      when "not-contains"
+        [{ "check_kind" => "not_contains", "target" => "response_text", "value" => text }, "Does not contain #{text.truncate(30).inspect}"]
+      when "not-icontains"
+        [{ "check_kind" => "not_contains", "target" => "response_text", "value" => text, "case_sensitive" => false }, "Does not contain #{text.truncate(30).inspect}"]
+      when "equals"
+        [{ "check_kind" => "equals", "target" => "response_text", "value" => text }, "Equals #{text.truncate(30).inspect}"]
+      when "regex"
+        [{ "check_kind" => "regex", "target" => "response_text", "pattern" => text }, "Matches /#{text.truncate(30)}/"]
+      when "is-json"
+        [{ "check_kind" => "valid_json", "target" => "response_text" }, "Valid JSON"]
+      end
+    end
+
+    def unique_name(base)
+      @used_names ||= Metric.pluck(:name).to_set
+      candidate = base
+      counter = 2
+      while @used_names.include?(candidate)
+        candidate = "#{base} (#{counter})"
+        counter += 1
+      end
+      @used_names << candidate
+      candidate
+    end
+  end
+end

data/app/services/completion_kit/run_comparison.rb

@@ -0,0 +1,13 @@
+module CompletionKit
+  module RunComparison
+    module_function
+
+    def result_change(left_passed, right_passed)
+      return nil if left_passed.nil? || right_passed.nil?
+      return "broke" if left_passed && !right_passed
+      return "fixed" if !left_passed && right_passed
+
+      "same"
+    end
+  end
+end

data/app/views/completion_kit/dashboard/_failing_checks_card.html.erb

@@ -0,0 +1,19 @@
+<div class="ck-card ck-stat-card ck-rise" id="ck-failing-checks-card" style="--rise-delay: 200ms;">
+  <p class="ck-kicker">Failing checks · last 7 days</p>
+  <div class="ck-stat-card__body">
+    <span class="ck-stat-card__count<%= failing_checks[:count].positive? ? ' is-danger' : ' is-clean' %>"><%= failing_checks[:count] %></span>
+  </div>
+
+  <% if failing_checks[:items].any? %>
+    <ul class="ck-failure-list">
+      <% failing_checks[:items].first(5).each do |item| %>
+        <li class="ck-failure-list__item">
+          <span class="ck-failure-list__surface ck-failure-list__surface--check">check</span>
+          <%= link_to item[:metric_name], completion_kit.run_path(item[:run]), class: "ck-link ck-failure-list__cause" %>
+        </li>
+      <% end %>
+    </ul>
+  <% else %>
+    <div class="ck-stat-card__foot"><span>No failing checks this week.</span></div>
+  <% end %>
+</div>

data/app/views/completion_kit/dashboard/show.html.erb

@@ -28,7 +28,7 @@
 </nav>
 
 <% if @activity %>
-  <div class="ck-grid ck-grid--cards ck-grid--cards-3 ck-pulse-grid">
+  <div class="ck-grid ck-grid--cards ck-grid--cards-4 ck-pulse-grid">
     <div class="ck-card ck-stat-card ck-rise" style="--rise-delay: 60ms;">
       <p class="ck-kicker">Activity · last 14 days</p>
       <% activity_max = @activity.map { |d| d[:count] }.max %>
@@ -49,6 +49,9 @@
 
     <%= render "completion_kit/dashboard/failures_card",
                failures: @failures, ignored_failures: @ignored_failures %>
+
+    <%= render "completion_kit/dashboard/failing_checks_card",
+               failing_checks: @failing_checks %>
   </div>
 
   <div class="ck-card ck-card--spaced ck-rise" style="--rise-delay: 240ms;">

data/app/views/completion_kit/imports/create.html.erb

@@ -0,0 +1,89 @@
+<ol class="ck-breadcrumb">
+  <li><%= link_to "Metrics", metrics_path %></li>
+  <li><%= link_to "Import from Promptfoo", import_path %></li>
+  <li>Summary</li>
+</ol>
+
+<section class="ck-page-header">
+  <div>
+    <h1 class="ck-title">Import summary</h1>
+    <p class="ck-lead">Here is what mapped into CompletionKit and what could not be translated.</p>
+  </div>
+  <div class="ck-actions">
+    <%= link_to "Import another", import_path, class: ck_button_classes(:light, variant: :outline) %>
+  </div>
+</section>
+
+<section class="ck-card ck-card--spaced">
+  <p class="ck-kicker">Prompts</p>
+  <% if @result.prompts[:created].any? %>
+    <ul class="ck-list ck-list--compact">
+      <% @result.prompts[:created].each do |name| %>
+        <li><%= link_to name, prompts_path, class: "ck-link" %></li>
+      <% end %>
+    </ul>
+  <% else %>
+    <p class="ck-meta-copy">No prompts imported.</p>
+  <% end %>
+  <% if @result.prompts[:skipped].any? %>
+    <p class="ck-kicker">Skipped prompts</p>
+    <ul class="ck-list ck-list--compact">
+      <% @result.prompts[:skipped].each do |item| %>
+        <li><code><%= item[:value] %></code>: <span class="ck-meta-copy"><%= item[:reason] %></span></li>
+      <% end %>
+    </ul>
+  <% end %>
+</section>
+
+<section class="ck-card ck-card--spaced">
+  <p class="ck-kicker">Dataset</p>
+  <% if @result.dataset[:created] %>
+    <p><%= link_to @result.dataset[:created], datasets_path, class: "ck-link" %>, <%= pluralize(@result.dataset[:rows], "row") %>. Columns: <%= @result.dataset[:columns].join(", ") %>.</p>
+  <% else %>
+    <p class="ck-meta-copy">Skipped: <%= @result.dataset[:skipped] %>.</p>
+  <% end %>
+</section>
+
+<section class="ck-card ck-card--spaced">
+  <p class="ck-kicker">Metrics</p>
+  <% if @result.metrics[:created].any? %>
+    <ul class="ck-list ck-list--compact">
+      <% @result.metrics[:created].each do |metric| %>
+        <li>
+          <%= link_to metric[:name], metrics_path, class: "ck-link" %>
+          <span class="ck-chip ck-chip--soft"><%= metric[:type] == "check" ? "Check" : "Judge" %></span>
+        </li>
+      <% end %>
+    </ul>
+  <% else %>
+    <p class="ck-meta-copy">No metrics imported.</p>
+  <% end %>
+  <% if @result.metrics[:skipped].any? %>
+    <p class="ck-kicker">Skipped asserts</p>
+    <ul class="ck-list ck-list--compact">
+      <% @result.metrics[:skipped].each do |item| %>
+        <li><code><%= item[:type] %></code>: <span class="ck-meta-copy"><%= item[:reason] %></span></li>
+      <% end %>
+    </ul>
+  <% end %>
+</section>
+
+<section class="ck-card ck-card--spaced">
+  <p class="ck-kicker">Providers</p>
+  <% if @result.providers[:matched].any? %>
+    <p class="ck-meta-copy">Matched to configured credentials:</p>
+    <div class="ck-chip-row">
+      <% @result.providers[:matched].each do |name| %>
+        <span class="ck-chip ck-chip--done"><%= name %></span>
+      <% end %>
+    </div>
+  <% end %>
+  <% if @result.providers[:unmatched].any? %>
+    <p class="ck-meta-copy">Referenced but not configured here. <%= link_to "Add a provider", provider_credentials_path, class: "ck-link" %> to run against these.</p>
+    <div class="ck-chip-row">
+      <% @result.providers[:unmatched].each do |name| %>
+        <span class="ck-chip ck-chip--warning"><%= name %></span>
+      <% end %>
+    </div>
+  <% end %>
+</section>

data/app/views/completion_kit/imports/new.html.erb

@@ -0,0 +1,31 @@
+<ol class="ck-breadcrumb">
+  <li><%= link_to "Metrics", metrics_path %></li>
+  <li>Import from Promptfoo</li>
+</ol>
+
+<section class="ck-page-header">
+  <div>
+    <h1 class="ck-title">Import from Promptfoo</h1>
+    <p class="ck-lead">Paste or upload a <code>promptfooconfig.yaml</code> to bring its prompts, test vars, and assertions into CompletionKit. This maps the eval and judge half of Promptfoo. Anything it cannot translate is reported back on the next page, never dropped silently.</p>
+  </div>
+</section>
+
+<%= form_with url: import_path, method: :post, multipart: true do |form| %>
+  <div class="ck-card ck-form-card">
+    <div class="ck-field">
+      <%= form.label :config, "Paste config", class: "ck-label" %>
+      <%= form.text_area :config, rows: 16, class: "ck-input ck-input--area ck-input--code", placeholder: "prompts:\n  - \"Summarize {{article}}.\"\ntests:\n  - vars: { article: \"...\" }\ndefaultTest:\n  assert:\n    - type: contains\n      value: \"summary\"" %>
+    </div>
+
+    <div class="ck-field">
+      <%= form.label :file, "Or upload a file", class: "ck-label" %>
+      <%= form.file_field :file, accept: ".yaml,.yml", class: "ck-input" %>
+      <p class="ck-field-hint">A <code>.yaml</code> file takes precedence over the pasted text.</p>
+    </div>
+
+    <div class="ck-actions">
+      <%= link_to "Cancel", metrics_path, class: ck_button_classes(:light, variant: :outline), tabindex: "0" %>
+      <%= form.submit "Import", class: ck_button_classes(:dark) %>
+    </div>
+  </div>
+<% end %>

data/app/views/completion_kit/metrics/index.html.erb

@@ -4,6 +4,7 @@
     <p class="ck-lead">Scoring dimensions used to evaluate each response. Each defines one thing to assess, by an LLM judge on a 1-5 scale or a deterministic check.</p>
   </div>
   <div class="ck-actions">
+    <%= link_to "Import from Promptfoo", import_path, class: ck_button_classes(:light, variant: :outline) %>
     <%= link_to "New metric", new_metric_path, class: ck_button_classes(:dark) %>
   </div>
 </section>

data/app/views/completion_kit/prompts/index.html.erb

@@ -54,10 +54,15 @@
           <% family_runs = CompletionKit::Run.where(prompt_id: prompt.family_versions.select(:id)).display_scoped %>
           <% current_version_runs = prompt.runs.display_scoped.includes(responses: :reviews) %>
           <% best_score = current_version_runs.map(&:avg_score).compact.max %>
+          <% best_pass_rate = current_version_runs.map(&:check_pass_rate).compact.max %>
           <td>
             <% if best_score %>
               <span class="<%= ck_badge_classes(ck_score_kind(best_score)) %>"><%= best_score %></span>
-            <% else %>
+            <% end %>
+            <% if best_pass_rate %>
+              <span class="<%= ck_badge_classes(ck_pass_rate_kind(best_pass_rate)) %>"><%= (best_pass_rate * 100).round %>%</span>
+            <% end %>
+            <% unless best_score || best_pass_rate %>
               <span class="ck-prompts-table__dim">—</span>
             <% end %>
           </td>

data/app/views/completion_kit/prompts/show.html.erb

@@ -64,7 +64,9 @@
       </thead>
       <tbody>
         <% versions.each do |v| %>
-          <% best_score = v.runs.display_scoped.map(&:avg_score).compact.max %>
+          <% scoped_runs = v.runs.display_scoped %>
+          <% best_score = scoped_runs.map(&:avg_score).compact.max %>
+          <% best_pass_rate = scoped_runs.map(&:check_pass_rate).compact.max %>
           <% pred = predecessor_of[v] %>
           <tr class="<%= "ck-results-table__row--active" if v.id == @prompt.id %>" onclick="window.location='<%= prompt_path(v) %>'" style="cursor: pointer;">
             <td>
@@ -86,7 +88,11 @@
             <td>
               <% if best_score %>
                 <span class="<%= ck_badge_classes(ck_score_kind(best_score)) %>"><%= best_score %></span>
-              <% else %>
+              <% end %>
+              <% if best_pass_rate %>
+                <span class="<%= ck_badge_classes(ck_pass_rate_kind(best_pass_rate)) %>"><%= (best_pass_rate * 100).round %>%</span>
+              <% end %>
+              <% unless best_score || best_pass_rate %>
                 <span class="ck-prompts-table__dim">—</span>
               <% end %>
             </td>

data/app/views/completion_kit/responses/show.html.erb

@@ -116,6 +116,8 @@
                 <% 5.times do |i| %>
                   <svg viewBox="0 0 24 24" width="16" height="16" stroke-width="1.75" class="ck-star <%= i < review.ai_score.to_i ? "ck-star--filled" : "ck-star--empty" %>"><polygon points="12 2 15.09 8.26 22 9.27 17 14.14 18.18 21.02 12 17.77 5.82 21.02 7 14.14 2 9.27 8.91 8.26 12 2"/></svg>
                 <% end %>
+              <% elsif review.check? && !review.passed.nil? %>
+                <%= ck_check_badge(review.passed) %>
               <% else %>
                 <span class="<%= ck_badge_classes(:pending) %>">Pending</span>
               <% end %>

data/app/views/completion_kit/runs/_response_row.html.erb

@@ -10,14 +10,21 @@
     <% end %>
   </td>
   <td data-label="Metrics">
-    <% scored_reviews = response.reviews.select { |r| r.ai_score.present? }.sort_by { |r| r.metric_name.to_s.downcase } %>
-    <% if scored_reviews.any? %>
+    <% pip_reviews = response.reviews.select { |r| r.ai_score.present? || !r.passed.nil? }.sort_by { |r| r.metric_name.to_s.downcase } %>
+    <% if pip_reviews.any? %>
       <span class="ck-metric-bar ck-metric-bar--compact">
-        <% scored_reviews.each do |r| %>
-          <span class="ck-metric-pip ck-metric-pip--<%= ck_score_kind(r.ai_score.to_f) %>">
-            <span class="ck-metric-pip__bar"></span>
-            <span class="ck-metric-pip__label"><%= r.metric_name %> <strong><%= r.ai_score %></strong></span>
-          </span>
+        <% pip_reviews.each do |r| %>
+          <% if r.ai_score.present? %>
+            <span class="ck-metric-pip ck-metric-pip--<%= ck_score_kind(r.ai_score.to_f) %>">
+              <span class="ck-metric-pip__bar"></span>
+              <span class="ck-metric-pip__label"><%= r.metric_name %> <strong><%= r.ai_score %></strong></span>
+            </span>
+          <% else %>
+            <span class="ck-metric-pip ck-metric-pip--<%= ck_pass_rate_kind(r.passed ? 1.0 : 0.0) %>">
+              <span class="ck-metric-pip__bar"></span>
+              <span class="ck-metric-pip__label"><%= r.metric_name %> <strong><%= r.passed ? "Pass" : "Fail" %></strong></span>
+            </span>
+          <% end %>
         <% end %>
       </span>
     <% else %>
@@ -26,7 +33,12 @@
   </td>
   <td data-label="Avg score">
     <% if response.reviewed? %>
-      <span class="<%= ck_badge_classes(ck_score_kind(response.score.to_f)) %>"><%= response.score %></span>
+      <% if response.score %>
+        <span class="<%= ck_badge_classes(ck_score_kind(response.score.to_f)) %>"><%= response.score %></span>
+      <% end %>
+      <% if response.checks_total.positive? %>
+        <span class="<%= ck_badge_classes(ck_pass_rate_kind(response.checks_passed.to_f / response.checks_total)) %>"><%= response.checks_passed %>/<%= response.checks_total %></span>
+      <% end %>
     <% else %>
       <span class="ck-response-cell__dim">—</span>
     <% end %>

data/app/views/completion_kit/runs/_row.html.erb

@@ -34,10 +34,17 @@
     <% if metrics.any? %>
       <div class="ck-metric-bar ck-metric-bar--compact">
         <% metrics.each do |m| %>
-          <span class="ck-metric-pip ck-metric-pip--<%= ck_score_kind(m[:avg]) %>">
-            <span class="ck-metric-pip__bar"></span>
-            <span class="ck-metric-pip__label"><%= m[:name] %> <strong><%= m[:avg] %></strong></span>
-          </span>
+          <% if m[:kind] == "check" %>
+            <span class="ck-metric-pip ck-metric-pip--<%= ck_pass_rate_kind(m[:pass_rate]) %>">
+              <span class="ck-metric-pip__bar"></span>
+              <span class="ck-metric-pip__label"><%= m[:name] %> <strong><%= (m[:pass_rate] * 100).round %>%</strong></span>
+            </span>
+          <% else %>
+            <span class="ck-metric-pip ck-metric-pip--<%= ck_score_kind(m[:avg]) %>">
+              <span class="ck-metric-pip__bar"></span>
+              <span class="ck-metric-pip__label"><%= m[:name] %> <strong><%= m[:avg] %></strong></span>
+            </span>
+          <% end %>
         <% end %>
       </div>
     <% else %>
@@ -46,8 +53,14 @@
   </td>
   <td>
     <% avg = run.avg_score %>
-    <% if avg %>
-      <span class="<%= ck_badge_classes(ck_score_kind(avg)) %>"><%= avg %></span>
+    <% pass_rate = run.check_pass_rate %>
+    <% if avg || pass_rate %>
+      <% if avg %>
+        <span class="<%= ck_badge_classes(ck_score_kind(avg)) %>"><%= avg %></span>
+      <% end %>
+      <% if pass_rate %>
+        <span class="<%= ck_badge_classes(ck_pass_rate_kind(pass_rate)) %>"><%= (pass_rate * 100).round %>%</span>
+      <% end %>
     <% else %>
       <span class="ck-runs-table__dim">—</span>
     <% end %>

data/app/views/completion_kit/runs/_sort_toolbar.html.erb

@@ -1,5 +1,5 @@
 <div id="run_sort_toolbar">
-  <% if run.judge_configured? %>
+  <% if run.gradable? %>
     <% active = run.status == "completed" && run.responses.joins(:reviews).exists? %>
     <div class="ck-toolbar" style="margin-top: 1.5rem;<%= ' visibility: hidden;' unless active %>" aria-hidden="<%= !active %>">
       <%= link_to "Best first", run_path(run, sort: "score_desc"), class: params[:sort].blank? || params[:sort] == "score_desc" ? ck_button_classes(:dark) : ck_button_classes(:light, variant: :outline), tabindex: active ? nil : -1 %>

data/app/views/completion_kit/runs/_status_panel.html.erb

@@ -2,7 +2,9 @@
 <div id="run_status_panel" aria-live="polite" aria-atomic="true">
   <% if run.status.in?(%w[running completed]) && snap[:generated_total] > 0 %>
     <% failed_count = snap[:generated_failed] + snap[:judged_failed] %>
-    <% has_judge = snap[:judged_total] > 0 || run.judge_configured? %>
+    <% show_grading = snap[:judged_total] > 0 || run.gradable? %>
+    <% has_llm_metric = run.llm_metrics.any? %>
+    <% has_checks = run.check_metrics.any? %>
     <% metric_avgs = run.metric_averages.sort_by { |m| m[:name].to_s.downcase } %>
     <% metric_lookup = metric_avgs.index_by { |m| m[:name].to_s.downcase } %>
     <section class="ck-run-status ck-run-status--<%= run.status %>">
@@ -18,7 +20,7 @@
           <p class="ck-run-status__cell-value ck-run-status__summary-line">
             <span class="ck-run-status__summary-num"><%= snap[:generated_done] %></span>
             <span class="ck-run-status__summary-text">of <%= snap[:generated_total] %> responses</span>
-            <% if has_judge %>
+            <% if show_grading %>
               <span class="ck-run-status__summary-sep">·</span>
               <span class="ck-run-status__summary-num"><%= snap[:judged_done] %></span>
               <span class="ck-run-status__summary-text">of <%= snap[:judged_total] %> judged</span>
@@ -30,7 +32,7 @@
           </p>
         </div>
 
-        <% if has_judge %>
+        <% if show_grading %>
           <div class="ck-run-status__cell">
             <p class="ck-run-status__metric-label">Metrics</p>
             <div class="ck-run-status__cell-value">
@@ -38,7 +40,12 @@
                 <div class="ck-metric-bar ck-metric-bar--compact">
                   <% run.metrics.order(:name).each do |metric| %>
                     <% avg_for_metric = metric_lookup[metric.name.to_s.downcase] %>
-                    <% if avg_for_metric %>
+                    <% if avg_for_metric && avg_for_metric[:kind] == "check" %>
+                      <span class="ck-metric-pip ck-metric-pip--<%= ck_pass_rate_kind(avg_for_metric[:pass_rate]) %>">
+                        <span class="ck-metric-pip__bar"></span>
+                        <span class="ck-metric-pip__label"><%= metric.name %> <strong><%= (avg_for_metric[:pass_rate] * 100).round %>%</strong></span>
+                      </span>
+                    <% elsif avg_for_metric %>
                       <span class="ck-metric-pip ck-metric-pip--<%= ck_score_kind(avg_for_metric[:avg]) %>">
                         <span class="ck-metric-pip__bar"></span>
                         <span class="ck-metric-pip__label"><%= metric.name %> <strong><%= avg_for_metric[:avg] %></strong></span>
@@ -57,16 +64,31 @@
             </div>
           </div>
 
-          <div class="ck-run-status__cell">
-            <p class="ck-run-status__metric-label">Avg score</p>
-            <div class="ck-run-status__cell-value">
-              <% if run.avg_score %>
-                <span class="<%= ck_badge_classes(ck_score_kind(run.avg_score)) %> ck-badge--lg"><%= run.avg_score %></span>
-              <% else %>
-                <span class="ck-run-status__cell-empty">—</span>
-              <% end %>
+          <% if has_llm_metric %>
+            <div class="ck-run-status__cell">
+              <p class="ck-run-status__metric-label">Avg score</p>
+              <div class="ck-run-status__cell-value">
+                <% if run.avg_score %>
+                  <span class="<%= ck_badge_classes(ck_score_kind(run.avg_score)) %> ck-badge--lg"><%= run.avg_score %></span>
+                <% else %>
+                  <span class="ck-run-status__cell-empty">—</span>
+                <% end %>
+              </div>
             </div>
-          </div>
+          <% end %>
+
+          <% if has_checks %>
+            <div class="ck-run-status__cell">
+              <p class="ck-run-status__metric-label">Checks passed</p>
+              <div class="ck-run-status__cell-value">
+                <% if run.check_pass_rate %>
+                  <span class="<%= ck_badge_classes(ck_pass_rate_kind(run.check_pass_rate)) %> ck-badge--lg"><%= (run.check_pass_rate * 100).round %>%</span>
+                <% else %>
+                  <span class="ck-run-status__cell-empty">—</span>
+                <% end %>
+              </div>
+            </div>
+          <% end %>
         <% end %>
       </div>
 

data/app/views/completion_kit/runs/compare.html.erb

@@ -46,21 +46,35 @@
             <% end %>
             <td><%= pm[:metric_name] %></td>
             <td>
-              <% if pm[:left_score] %>
+              <% if pm[:kind] == "check" %>
+                <% if pm[:left_passed].nil? %>
+                  <span class="ck-meta-copy">—</span>
+                <% else %>
+                  <%= ck_check_badge(pm[:left_passed]) %>
+                <% end %>
+              <% elsif pm[:left_score] %>
                 <span class="<%= ck_badge_classes(ck_score_kind(pm[:left_score].to_f)) %>"><%= pm[:left_score] %></span>
               <% else %>
                 <span class="ck-meta-copy">—</span>
               <% end %>
             </td>
             <td>
-              <% if pm[:right_score] %>
+              <% if pm[:kind] == "check" %>
+                <% if pm[:right_passed].nil? %>
+                  <span class="ck-meta-copy">—</span>
+                <% else %>
+                  <%= ck_check_badge(pm[:right_passed]) %>
+                <% end %>
+              <% elsif pm[:right_score] %>
                 <span class="<%= ck_badge_classes(ck_score_kind(pm[:right_score].to_f)) %>"><%= pm[:right_score] %></span>
               <% else %>
                 <span class="ck-meta-copy">—</span>
               <% end %>
             </td>
             <td>
-              <% if pm[:delta] %>
+              <% if pm[:kind] == "check" %>
+                <%= ck_result_change_badge(pm[:result_change]) || content_tag(:span, "—", class: "ck-meta-copy") %>
+              <% elsif pm[:delta] %>
                 <% delta_class = pm[:delta] > 0 ? "ck-delta--positive" : pm[:delta] < 0 ? "ck-delta--negative" : "ck-delta--zero" %>
                 <span class="ck-delta <%= delta_class %>"><%= pm[:delta].positive? ? "+#{pm[:delta]}" : pm[:delta].to_s %></span>
               <% else %>

data/app/views/completion_kit/suggestions/show.html.erb

@@ -20,6 +20,9 @@
       <% if @run.avg_score %>
         <span class="<%= ck_badge_classes(ck_score_kind(@run.avg_score)) %>"><%= @run.avg_score %></span>
       <% end %>
+      <% if @run.check_pass_rate %>
+        <span class="<%= ck_badge_classes(ck_pass_rate_kind(@run.check_pass_rate)) %>"><%= (@run.check_pass_rate * 100).round %>%</span>
+      <% end %>
     </p>
   </div>
   <div class="ck-actions">

data/config/routes.rb

@@ -59,6 +59,9 @@ CompletionKit::Engine.routes.draw do
 
   get "api_reference", to: "api_reference#index", as: :api_reference
 
+  get "import", to: "imports#new", as: :import
+  post "import", to: "imports#create"
+
   namespace :api do
     namespace :v1 do
       resources :prompts do
@@ -95,6 +98,7 @@ CompletionKit::Engine.routes.draw do
       resources :tags
       resources :provider_credentials
       resources :agreements, only: [:index, :destroy]
+      post "imports/promptfoo", to: "imports#promptfoo"
     end
   end
 

data/lib/completion_kit/version.rb

@@ -1,3 +1,3 @@
 module CompletionKit
-  VERSION = "0.18.1"
+  VERSION = "0.20.0"
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: completion-kit
 version: !ruby/object:Gem::Version
-  version: 0.18.1
+  version: 0.20.0
 platform: ruby
 authors:
 - Damien Bastin
@@ -238,6 +238,7 @@ files:
 - app/controllers/completion_kit/api/v1/agreements_controller.rb
 - app/controllers/completion_kit/api/v1/base_controller.rb
 - app/controllers/completion_kit/api/v1/datasets_controller.rb
+- app/controllers/completion_kit/api/v1/imports_controller.rb
 - app/controllers/completion_kit/api/v1/metric_groups_controller.rb
 - app/controllers/completion_kit/api/v1/metric_versions_controller.rb
 - app/controllers/completion_kit/api/v1/metrics_controller.rb
@@ -251,6 +252,7 @@ files:
 - app/controllers/completion_kit/dashboard_controller.rb
 - app/controllers/completion_kit/dashboard_dismissals_controller.rb
 - app/controllers/completion_kit/datasets_controller.rb
+- app/controllers/completion_kit/imports_controller.rb
 - app/controllers/completion_kit/mcp_controller.rb
 - app/controllers/completion_kit/metric_groups_controller.rb
 - app/controllers/completion_kit/metrics_controller.rb
@@ -261,6 +263,7 @@ files:
 - app/controllers/completion_kit/runs_controller.rb
 - app/controllers/completion_kit/suggestions_controller.rb
 - app/controllers/completion_kit/tags_controller.rb
+- app/controllers/concerns/completion_kit/response_ordering.rb
 - app/controllers/concerns/completion_kit/tag_filtering.rb
 - app/helpers/completion_kit/application_helper.rb
 - app/jobs/completion_kit/application_job.rb
@@ -316,6 +319,7 @@ files:
 - app/services/completion_kit/mcp_tools/agreements.rb
 - app/services/completion_kit/mcp_tools/base.rb
 - app/services/completion_kit/mcp_tools/datasets.rb
+- app/services/completion_kit/mcp_tools/imports.rb
 - app/services/completion_kit/mcp_tools/judges.rb
 - app/services/completion_kit/mcp_tools/metric_groups.rb
 - app/services/completion_kit/mcp_tools/metric_versions.rb
@@ -338,7 +342,9 @@ files:
 - app/services/completion_kit/open_router_client.rb
 - app/services/completion_kit/prompt_improvement_service.rb
 - app/services/completion_kit/prompt_improvement_validator.rb
+- app/services/completion_kit/promptfoo_importer.rb
 - app/services/completion_kit/provider_endpoint.rb
+- app/services/completion_kit/run_comparison.rb
 - app/services/completion_kit/starter_metrics.rb
 - app/services/completion_kit/worker_health.rb
 - app/validators/completion_kit/tenant_scoped_uniqueness_validator.rb
@@ -352,6 +358,7 @@ files:
 - app/views/completion_kit/api_reference/index.html.erb
 - app/views/completion_kit/dashboard/_eye_icon.html.erb
 - app/views/completion_kit/dashboard/_eye_off_icon.html.erb
+- app/views/completion_kit/dashboard/_failing_checks_card.html.erb
 - app/views/completion_kit/dashboard/_failures_card.html.erb
 - app/views/completion_kit/dashboard/_worst_metric_card.html.erb
 - app/views/completion_kit/dashboard/show.html.erb
@@ -361,6 +368,8 @@ files:
 - app/views/completion_kit/datasets/index.html.erb
 - app/views/completion_kit/datasets/new.html.erb
 - app/views/completion_kit/datasets/show.html.erb
+- app/views/completion_kit/imports/create.html.erb
+- app/views/completion_kit/imports/new.html.erb
 - app/views/completion_kit/metric_groups/_form.html.erb
 - app/views/completion_kit/metric_groups/edit.html.erb
 - app/views/completion_kit/metric_groups/index.html.erb