decision_agent 0.3.0 → 1.1.0

data/spec/agent_spec.rb

@@ -1,289 +0,0 @@
-require "spec_helper"
-
-RSpec.describe DecisionAgent::Agent do
-  describe "#initialize" do
-    it "requires at least one evaluator" do
-      expect do
-        DecisionAgent::Agent.new(evaluators: [])
-      end.to raise_error(DecisionAgent::InvalidConfigurationError, /at least one evaluator/i)
-    end
-
-    it "validates evaluators respond to #evaluate" do
-      invalid_evaluator = Object.new
-
-      expect do
-        DecisionAgent::Agent.new(evaluators: [invalid_evaluator])
-      end.to raise_error(DecisionAgent::InvalidEvaluatorError)
-    end
-
-    it "validates scoring strategy responds to #score" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-      invalid_strategy = Object.new
-
-      expect do
-        DecisionAgent::Agent.new(
-          evaluators: [evaluator],
-          scoring_strategy: invalid_strategy
-        )
-      end.to raise_error(DecisionAgent::InvalidScoringStrategyError)
-    end
-
-    it "validates audit adapter responds to #record" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-      invalid_adapter = Object.new
-
-      expect do
-        DecisionAgent::Agent.new(
-          evaluators: [evaluator],
-          audit_adapter: invalid_adapter
-        )
-      end.to raise_error(DecisionAgent::InvalidAuditAdapterError)
-    end
-
-    it "uses defaults when optional parameters are omitted" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-
-      agent = DecisionAgent::Agent.new(evaluators: [evaluator])
-
-      expect(agent.scoring_strategy).to be_a(DecisionAgent::Scoring::WeightedAverage)
-      expect(agent.audit_adapter).to be_a(DecisionAgent::Audit::NullAdapter)
-    end
-
-    it "enables validation by default in non-production environments" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-      original_env = ENV.fetch("RAILS_ENV", nil)
-      ENV["RAILS_ENV"] = "development"
-
-      agent = DecisionAgent::Agent.new(evaluators: [evaluator])
-      # Validation should be enabled (we can't directly test this, but we can test behavior)
-      # If validation is enabled, invalid evaluations would raise errors
-      expect(agent).to be_a(DecisionAgent::Agent)
-
-      ENV["RAILS_ENV"] = original_env
-    end
-
-    it "disables validation in production by default" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-      original_env = ENV.fetch("RAILS_ENV", nil)
-      ENV["RAILS_ENV"] = "production"
-
-      agent = DecisionAgent::Agent.new(evaluators: [evaluator])
-      expect(agent).to be_a(DecisionAgent::Agent)
-
-      ENV["RAILS_ENV"] = original_env
-    end
-
-    it "allows explicit validation control" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-
-      agent_with_validation = DecisionAgent::Agent.new(
-        evaluators: [evaluator],
-        validate_evaluations: true
-      )
-      expect(agent_with_validation).to be_a(DecisionAgent::Agent)
-
-      agent_without_validation = DecisionAgent::Agent.new(
-        evaluators: [evaluator],
-        validate_evaluations: false
-      )
-      expect(agent_without_validation).to be_a(DecisionAgent::Agent)
-    end
-  end
-
-  describe "#decide" do
-    it "returns a Decision object with all required fields" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(
-        decision: "approve",
-        weight: 0.8,
-        reason: "Test approval"
-      )
-
-      agent = DecisionAgent::Agent.new(evaluators: [evaluator])
-
-      result = agent.decide(context: { user: "test" })
-
-      expect(result).to be_a(DecisionAgent::Decision)
-      expect(result.decision).to eq("approve")
-      expect(result.confidence).to be_between(0.0, 1.0)
-      expect(result.explanations).to be_an(Array)
-      expect(result.evaluations).to be_an(Array)
-      expect(result.audit_payload).to be_a(Hash)
-    end
-
-    it "accepts Context object or Hash for context parameter" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-      agent = DecisionAgent::Agent.new(evaluators: [evaluator])
-
-      result1 = agent.decide(context: { user: "test" })
-      result2 = agent.decide(context: DecisionAgent::Context.new({ user: "test" }))
-
-      expect(result1.decision).to eq(result2.decision)
-    end
-
-    it "raises NoEvaluationsError when no evaluators return decisions" do
-      failing_evaluator = Class.new(DecisionAgent::Evaluators::Base) do
-        def evaluate(_context, feedback: {})
-          nil
-        end
-      end
-
-      agent = DecisionAgent::Agent.new(evaluators: [failing_evaluator.new])
-
-      expect do
-        agent.decide(context: {})
-      end.to raise_error(DecisionAgent::NoEvaluationsError)
-    end
-
-    it "includes feedback in evaluation" do
-      feedback_evaluator = Class.new(DecisionAgent::Evaluators::Base) do
-        def evaluate(_context, feedback: {})
-          decision = feedback[:override] ? "reject" : "approve"
-          DecisionAgent::Evaluation.new(
-            decision: decision,
-            weight: 1.0,
-            reason: "Feedback-based",
-            evaluator_name: "FeedbackEvaluator"
-          )
-        end
-      end
-
-      agent = DecisionAgent::Agent.new(evaluators: [feedback_evaluator.new])
-
-      result1 = agent.decide(context: {}, feedback: {})
-      result2 = agent.decide(context: {}, feedback: { override: true })
-
-      expect(result1.decision).to eq("approve")
-      expect(result2.decision).to eq("reject")
-    end
-
-    it "records decision via audit adapter" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-
-      audit_adapter = Class.new(DecisionAgent::Audit::Adapter) do
-        attr_reader :recorded_decision, :recorded_context
-
-        def record(decision, context)
-          @recorded_decision = decision
-          @recorded_context = context
-        end
-      end.new
-
-      agent = DecisionAgent::Agent.new(
-        evaluators: [evaluator],
-        audit_adapter: audit_adapter
-      )
-
-      result = agent.decide(context: { user: "test" })
-
-      expect(audit_adapter.recorded_decision).to eq(result)
-      expect(audit_adapter.recorded_context.to_h).to eq({ user: "test" })
-    end
-
-    it "includes deterministic hash in audit payload" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve", weight: 0.8)
-      agent = DecisionAgent::Agent.new(evaluators: [evaluator])
-
-      result1 = agent.decide(context: { user: "test" })
-      result2 = agent.decide(context: { user: "test" })
-
-      expect(result1.audit_payload[:deterministic_hash]).to be_a(String)
-      expect(result1.audit_payload[:deterministic_hash]).to eq(result2.audit_payload[:deterministic_hash])
-    end
-
-    it "produces different hashes for different contexts" do
-      evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-      agent = DecisionAgent::Agent.new(evaluators: [evaluator])
-
-      result1 = agent.decide(context: { user: "alice" })
-      result2 = agent.decide(context: { user: "bob" })
-
-      expect(result1.audit_payload[:deterministic_hash]).not_to eq(result2.audit_payload[:deterministic_hash])
-    end
-  end
-
-  describe "conflict resolution" do
-    it "resolves conflicting evaluations using scoring strategy" do
-      evaluator1 = DecisionAgent::Evaluators::StaticEvaluator.new(
-        decision: "approve",
-        weight: 0.6,
-        name: "Evaluator1"
-      )
-      evaluator2 = DecisionAgent::Evaluators::StaticEvaluator.new(
-        decision: "reject",
-        weight: 0.9,
-        name: "Evaluator2"
-      )
-
-      agent = DecisionAgent::Agent.new(
-        evaluators: [evaluator1, evaluator2],
-        scoring_strategy: DecisionAgent::Scoring::MaxWeight.new
-      )
-
-      result = agent.decide(context: {})
-
-      expect(result.decision).to eq("reject")
-      expect(result.explanations.join(" ")).to include("Conflicting evaluations")
-    end
-
-    it "includes conflicting evaluations in explanations" do
-      evaluator1 = DecisionAgent::Evaluators::StaticEvaluator.new(
-        decision: "approve",
-        weight: 0.4,
-        name: "Evaluator1"
-      )
-      evaluator2 = DecisionAgent::Evaluators::StaticEvaluator.new(
-        decision: "reject",
-        weight: 0.7,
-        name: "Evaluator2"
-      )
-
-      agent = DecisionAgent::Agent.new(evaluators: [evaluator1, evaluator2])
-
-      result = agent.decide(context: {})
-
-      explanations_text = result.explanations.join(" ")
-      expect(explanations_text).to include("Evaluator1")
-      expect(explanations_text).to include("Evaluator2")
-    end
-  end
-
-  describe "multiple evaluators agreeing" do
-    it "combines evaluations when all agree" do
-      evaluator1 = DecisionAgent::Evaluators::StaticEvaluator.new(
-        decision: "approve",
-        weight: 0.6,
-        name: "Evaluator1"
-      )
-      evaluator2 = DecisionAgent::Evaluators::StaticEvaluator.new(
-        decision: "approve",
-        weight: 0.8,
-        name: "Evaluator2"
-      )
-
-      agent = DecisionAgent::Agent.new(evaluators: [evaluator1, evaluator2])
-
-      result = agent.decide(context: {})
-
-      expect(result.decision).to eq("approve")
-      expect(result.confidence).to be > 0.5
-    end
-  end
-
-  describe "graceful error handling" do
-    it "ignores evaluators that raise errors" do
-      good_evaluator = DecisionAgent::Evaluators::StaticEvaluator.new(decision: "approve")
-
-      bad_evaluator = Class.new(DecisionAgent::Evaluators::Base) do
-        def evaluate(_context, feedback: {})
-          raise StandardError, "Intentional error"
-        end
-      end
-
-      agent = DecisionAgent::Agent.new(evaluators: [bad_evaluator.new, good_evaluator])
-
-      result = agent.decide(context: {})
-
-      expect(result.decision).to eq("approve")
-    end
-  end
-end