fix: centralize AI model configuration and fix invalid model name

[jdy8739]

• Replace invalid "gpt-5.2" with valid "gpt-4o-mini" model
• Create AI_MODEL constant as single source of truth in features/ai/lib/agent.ts
• Update all AI features (correction, story generation) to use centralized agent.model
• Remove duplicate OpenAI client initializations across files
• Ensures consistent model usage across all AI features

[claude]

Pull Request Review

✅ Overall Assessment

This is an excellent refactoring that improves code quality, maintainability, and follows best practices. The PR successfully centralizes AI model configuration and fixes a critical bug.

---

🎯 Positive Points

1. Critical Bug Fix

• ✅ Fixed invalid model name "gpt-5.2" → "gpt-4o-mini" (GPT-5.2 doesn't exist)
• This would have caused runtime errors when the agent was used

2. Single Source of Truth Pattern

• ✅ Created AI_MODEL constant in features/ai/lib/agent.ts
• ✅ Eliminates duplicate model strings across files
• ✅ Makes future model changes trivial (update one line)

3. Code Deduplication

• ✅ Removed duplicate OpenAI client initializations
• ✅ Centralized openai and agent exports
• Files affected:
  • app/api/ai/story/route.ts (-7 lines, +2 lines)
  • features/story/lib/actions.ts (-7 lines, +2 lines)

4. Adherence to Project Architecture

• ✅ Follows Feature-Sliced Design (FSD) principles
• ✅ Proper use of feature modules (features/ai/lib/agent.ts)
• ✅ Shared AI functionality correctly placed in feature module

---

🔍 Code Quality Analysis

Type Safety

const AI_MODEL = "gpt-4o-mini" as const;

✅ Excellent: Uses as const assertion for type safety (matches CLAUDE.md guidelines)

Import Consistency

import { openai, agent } from "@/features/ai/lib/agent";

✅ Good: Uses barrel exports and path aliases

Environment Variables

• ✅ No .env file modifications (complies with critical policy in CLAUDE.md)
• ✅ Uses process.env.OPENAI_API_KEY consistently

---

🔒 Security Considerations

✅ No security concerns - This is purely a refactoring PR:

• No new API endpoints
• No authentication changes
• No data exposure risks
• Environment variables handled correctly

---

🚀 Performance Considerations

✅ Neutral to Positive:

• Eliminates duplicate OpenAI client instantiations
• Reuses single client instance across modules
• No negative performance impact

---

🧪 Test Coverage

⚠️ Minor Observation:

• No test files modified or added
• Recommendation: Consider adding a test to verify agent.model equals "gpt-4o-mini"
• Example:

  // features/ai/lib/__tests__/agent.test.ts
  import { agent } from '../agent';
  
  test('agent uses correct model', () => {
    expect(agent.model).toBe('gpt-4o-mini');
  });

---

📝 Documentation & Maintainability

✅ Excellent:

• Clear PR description explains the "why" and "what"
• Inline comment: "// SINGLE SOURCE OF TRUTH: AI Model Configuration" is helpful
• Comment explains impact: "// Change this one constant to update all AI features across the app"

---

🎨 Code Style & Conventions

✅ Fully compliant with CLAUDE.md rules:

• Arrow functions not applicable (constants only)
• Exports at bottom: ✅ export { openai, agent };
• Const assertions: ✅ as const
• No default exports used
• Clean barrel exports pattern

---

🐛 Potential Issues

None found. The code is clean and well-structured.

---

💡 Suggestions for Future Improvements (Optional)

Click to expand suggestions

1. Environment Validation Enhancement

Consider adding runtime validation that the model name is valid (though this is not critical):

const VALID_MODELS = ['gpt-4o-mini', 'gpt-4o', 'gpt-4-turbo'] as const;
type ValidModel = typeof VALID_MODELS[number];
const AI_MODEL: ValidModel = "gpt-4o-mini";

2. Model Configuration Object

For future scalability, consider expanding to a configuration object:

const AI_CONFIG = {
  model: "gpt-4o-mini",
  temperature: 0.7,
  maxTokens: 300,
} as const;

3. Type Export

Consider exporting the agent type for better TypeScript support:

export type Agent = typeof agent;

---

✅ Final Recommendation

APPROVE AND MERGE ✅

This PR:

• ✅ Fixes a critical bug (invalid model name)
• ✅ Improves code quality through DRY principles
• ✅ Follows all project conventions from CLAUDE.md
• ✅ Has no security or performance concerns
• ✅ Makes future changes easier

Impact: Low risk, high value refactoring.

---

📊 Metrics

• Files changed: 3
• Additions: +9 lines
• Deletions: -15 lines
• Net change: -6 lines (cleaner codebase!)
• Complexity: Reduced (DRY principle applied)

Great work! 🎉