activeagent 1.8.0 → 1.8.1

data/lib/active_agent/evals/result.rb

@@ -49,6 +49,37 @@ module ActiveAgent
         diagnosis&.dig("judge", "suggested_tool")
       end
 
+      # Where the replay's cost came from, as the caller recorded it in the
+      # replay metadata: "reported" (the caller priced it), "estimated"
+      # (tokens × a model rate), "no_usage" (nothing was generated, so $0.00)
+      # or "unpriced" (nothing to price it from). A cost with no record is
+      # the caller's own figure; no cost at all is unpriced.
+      def cost_source
+        recorded = replay_meta("cost_source").to_s
+        return recorded if recorded.present?
+
+        replay.cost.nil? ? "unpriced" : "reported"
+      end
+
+      def estimated_cost?
+        cost_source == "estimated"
+      end
+
+      # The rate an estimated cost was worked out at, `{ "input", "output",
+      # "source" }` in $ per million tokens, or nil.
+      def cost_rate
+        rate = replay_meta("cost_rate")
+        rate.is_a?(Hash) ? rate.transform_keys(&:to_s) : nil
+      end
+
+      # What the judge spent on this result — `{ "calls", "input_tokens",
+      # "output_tokens", "cost", "model", "by_kind", "source" }` — or nil
+      # when no judge was asked about it.
+      def judge_usage
+        usage = replay_meta("judge_usage")
+        usage.is_a?(Hash) ? usage.transform_keys(&:to_s) : nil
+      end
+
       def to_h
         {
           "scenario_key" => scenario.key,
@@ -70,9 +101,18 @@ module ActiveAgent
           "fault" => fault,
           "recommendation" => recommendation,
           "diagnosis" => diagnosis,
+          "judge_usage" => judge_usage,
           "metadata" => replay.metadata
         }.compact
       end
+
+      private
+
+      # A caller may key the replay metadata with symbols.
+      def replay_meta(key)
+        metadata = replay.metadata
+        metadata[key] || metadata[key.to_sym]
+      end
     end
   end
 end

data/lib/active_agent/evals/runner.rb

@@ -51,10 +51,13 @@ module ActiveAgent
       # @param require_judge_scores [Boolean] fail an otherwise passing result
       #   when a requested task/LLM grade is unavailable, rather than falling
       #   back to rule scores. Does not require a judge for rules-only runs.
+      # @param release [Hash, nil] the release of the agent under evaluation,
+      #   `{ "digest", "revision", "label" }`, carried onto the Report so a
+      #   published run names the code it scored (see Report.new)
       def initialize(scenarios:, models:, replay:, criteria: [], judge: nil, judge_task: true, available_tools: {},
                      instructions: nil, agent_name: "The agent", threshold: PASS_THRESHOLD,
                      refine_faults: DEFAULT_REFINE_FAULTS, judge_limit: DEFAULT_JUDGE_LIMIT, on_result: nil,
-                     around_evaluation: nil, require_judge_scores: false, metadata: {})
+                     around_evaluation: nil, require_judge_scores: false, metadata: {}, release: nil)
         @scenarios = scenarios
         @models = models
         @replay = replay
@@ -71,6 +74,7 @@ module ActiveAgent
         @around_evaluation = around_evaluation
         @require_judge_scores = require_judge_scores
         @metadata = metadata
+        @release = release
         @judge_calls = 0
         @scorer = Scorer.new(criteria: criteria, judge: judge)
       end
@@ -83,7 +87,7 @@ module ActiveAgent
         end
 
         Report.new(results: results, models: @models, judge: @judge, instructions: @instructions,
-                   threshold: @threshold, metadata: @metadata)
+                   threshold: @threshold, metadata: @metadata, release: @release)
       end
 
       # Scores and diagnoses one replay. Public so a caller that has already run

data/lib/active_agent/evals.rb

@@ -17,6 +17,7 @@ require_relative "evals/scenario_parser"
 require_relative "evals/suite"
 require_relative "evals/model_spec"
 require_relative "evals/replay"
+require_relative "evals/format"
 require_relative "evals/scorer"
 require_relative "evals/diagnosis"
 require_relative "evals/judge"

data/lib/active_agent/version.rb

@@ -1,3 +1,3 @@
 module ActiveAgent
-  VERSION = "1.8.0"
+  VERSION = "1.8.1"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: activeagent
 version: !ruby/object:Gem::Version
-  version: 1.8.0
+  version: 1.8.1
 platform: ruby
 authors:
 - Justin Bowen
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-09-30 00:00:00.000000000 Z
+date: 2026-10-01 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: actionpack
@@ -443,6 +443,7 @@ files:
 - lib/active_agent/evals/correlation.rb
 - lib/active_agent/evals/design_tokens.rb
 - lib/active_agent/evals/diagnosis.rb
+- lib/active_agent/evals/format.rb
 - lib/active_agent/evals/judge.rb
 - lib/active_agent/evals/model_spec.rb
 - lib/active_agent/evals/publisher.rb