fix: enforce expected_status in test execution pipeline

[mikolajbadyl]

The agent was always comparing results against status 200, even when testing error cases (400, 404).

• agent.go — local test plan generator now includes expected_status in the JSON format and knows which status codes to use per method/scenario
• chat.go — added expected_status to ExecuteTestGroup tool schema (required field) with clear instructions
• update_tests.go — expected_status is now included in test cases returned to the agent after GenerateTestPlan

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Files Reviewed (7 files)

• internal/agents/agent.go - System prompt updates to require expected_status
• internal/agents/chat.go - Tool schema updates making expected_status required
• internal/agents/types.go - Example value update (200→404)
• internal/cli/handlers.go - Validation and execution logic
• internal/cli/update.go - Test case construction with expected_status
• internal/cli/update_tests.go - Test generation with expected_status
• internal/llm/claude/client.go - Schema handling refactor

---

Review Notes

This PR introduces a required expected_status field for test cases, which is a significant improvement for test accuracy. The changes ensure that:

1. Agent prompts now instruct the LLM to include expected_status for each test
2. Validation in ExecuteTestGroup rejects tests missing expected_status
3. Test execution uses the provided status code instead of defaulting to 200
4. Claude client was refactored to preserve full JSON schema structure

The flow is consistent: validation prevents tests without expected_status from proceeding, and the agent is informed to re-call with the correct field.