feat(agents): implement agent framework and CLI

tests/test_models.py

@@ -0,0 +1,224 @@
+"""Tests for data models."""
+
+from pathlib import Path
+
+import pytest
+
+from arbiter.models import (
+    AgentConfig,
+    AgentName,
+    Finding,
+    Policy,
+    ReviewResult,
+    Severity,
+    Verdict,
+)
+
+
+class TestEnums:
+    def test_severity_values(self) -> None:
+        assert Severity.CRITICAL == "critical"
+        assert Severity.HIGH == "high"
+        assert Severity.MEDIUM == "medium"
+        assert Severity.LOW == "low"
+        assert Severity.INFO == "info"
+
+    def test_verdict_values(self) -> None:
+        assert Verdict.APPROVE == "approve"
+        assert Verdict.REQUEST_CHANGES == "request_changes"
+        assert Verdict.COMMENT == "comment"
+
+    def test_agent_name_values(self) -> None:
+        assert AgentName.SECURITY == "security"
+        assert AgentName.STYLE == "style"
+        assert AgentName.COMPLEXITY == "complexity"
+
+    def test_severity_from_string(self) -> None:
+        assert Severity("critical") == Severity.CRITICAL
+        assert Severity("high") == Severity.HIGH
+
+
+class TestFinding:
+    def test_finding_creation(self) -> None:
+        finding = Finding(
+            id="test-123",
+            agent=AgentName.SECURITY,
+            file="src/auth.py",
+            line_start=10,
+            line_end=15,
+            severity=Severity.HIGH,
+            confidence=0.9,
+            title="SQL Injection",
+            description="User input concatenated in SQL query",
+            reasoning="Allows attackers to execute arbitrary SQL",
+            suggestion="Use parameterized queries",
+            references=["https://owasp.org"],
+            prompt_version="security-v1.0",
+        )
+        assert finding.id == "test-123"
+        assert finding.agent == AgentName.SECURITY
+        assert finding.severity == Severity.HIGH
+        assert finding.confidence == 0.9
+
+    def test_finding_confidence_validation(self) -> None:
+        with pytest.raises(ValueError):
+            Finding(
+                id="test",
+                agent=AgentName.SECURITY,
+                file="test.py",
+                line_start=1,
+                line_end=1,
+                severity=Severity.INFO,
+                confidence=1.5,
+                title="Test",
+                description="Test",
+                reasoning="Test",
+                prompt_version="test-v1.0",
+            )
+
+    def test_finding_line_validation(self) -> None:
+        with pytest.raises(ValueError):
+            Finding(
+                id="test",
+                agent=AgentName.SECURITY,
+                file="test.py",
+                line_start=0,
+                line_end=1,
+                severity=Severity.INFO,
+                confidence=0.5,
+                title="Test",
+                description="Test",
+                reasoning="Test",
+                prompt_version="test-v1.0",
+            )
+
+    def test_finding_serialization(self) -> None:
+        finding = Finding(
+            id="test-123",
+            agent=AgentName.SECURITY,
+            file="src/auth.py",
+            line_start=10,
+            line_end=15,
+            severity=Severity.HIGH,
+            confidence=0.9,
+            title="Test",
+            description="Test desc",
+            reasoning="Test reason",
+            prompt_version="security-v1.0",
+        )
+        data = finding.model_dump()
+        assert data["id"] == "test-123"
+        assert data["agent"] == "security"
+        assert data["severity"] == "high"
+
+
+class TestAgentConfig:
+    def test_default_values(self) -> None:
+        config = AgentConfig()
+        assert config.enabled is True
+        assert config.model is None
+        assert config.severity_threshold == Severity.INFO
+        assert config.prompt_additions is None
+
+    def test_custom_values(self) -> None:
+        config = AgentConfig(
+            enabled=False,
+            model="gpt-4o",
+            severity_threshold=Severity.MEDIUM,
+            prompt_additions="Focus on auth",
+        )
+        assert config.enabled is False
+        assert config.model == "gpt-4o"
+        assert config.severity_threshold == Severity.MEDIUM
+
+
+class TestPolicy:
+    def test_default_policy(self) -> None:
+        policy = Policy()
+        assert len(policy.agents) == 3
+        assert AgentName.SECURITY in policy.agents
+        assert AgentName.STYLE in policy.agents
+        assert AgentName.COMPLEXITY in policy.agents
+
+    def test_get_enabled_agents(self) -> None:
+        policy = Policy()
+        enabled = policy.get_enabled_agents()
+        assert len(enabled) == 3
+        assert AgentName.SECURITY in enabled
+
+    def test_get_enabled_agents_with_disabled(self) -> None:
+        policy = Policy(
+            agents={
+                AgentName.SECURITY: AgentConfig(enabled=True),
+                AgentName.STYLE: AgentConfig(enabled=False),
+                AgentName.COMPLEXITY: AgentConfig(enabled=True),
+            }
+        )
+        enabled = policy.get_enabled_agents()
+        assert len(enabled) == 2
+        assert AgentName.STYLE not in enabled
+
+    def test_load_from_yaml(self, tmp_path: Path) -> None:
+        policy_file = tmp_path / "policy.yaml"
+        policy_file.write_text("""
+version: "1.1"
+agents:
+  security:
+    enabled: true
+    model: gpt-4o
+    severity_threshold: high
+  style:
+    enabled: false
+  complexity:
+    enabled: true
+""")
+        policy = Policy.load(policy_file)
+        assert policy.version == "1.1"
+        assert policy.agents[AgentName.SECURITY].model == "gpt-4o"
+        assert policy.agents[AgentName.SECURITY].severity_threshold == Severity.HIGH
+        assert policy.agents[AgentName.STYLE].enabled is False
+
+    def test_load_empty_yaml(self, tmp_path: Path) -> None:
+        policy_file = tmp_path / "policy.yaml"
+        policy_file.write_text("")
+        policy = Policy.load(policy_file)
+        assert policy.version == "1.0"
+        assert len(policy.agents) == 3
+
+
+class TestReviewResult:
+    def test_review_result_creation(self) -> None:
+        result = ReviewResult(
+            agent_name=AgentName.SECURITY,
+            findings=[],
+            duration_ms=1000,
+            tokens_used=500,
+            cost_usd=0.01,
+        )
+        assert result.agent_name == AgentName.SECURITY
+        assert result.duration_ms == 1000
+        assert result.cost_usd == 0.01
+
+    def test_review_result_with_findings(self) -> None:
+        finding = Finding(
+            id="test-123",
+            agent=AgentName.SECURITY,
+            file="test.py",
+            line_start=1,
+            line_end=1,
+            severity=Severity.HIGH,
+            confidence=0.9,
+            title="Test",
+            description="Test",
+            reasoning="Test",
+            prompt_version="test-v1.0",
+        )
+        result = ReviewResult(
+            agent_name=AgentName.SECURITY,
+            findings=[finding],
+            duration_ms=1000,
+            tokens_used=500,
+            cost_usd=0.01,
+        )
+        assert len(result.findings) == 1
+        assert result.findings[0].id == "test-123"