Fix Ultrathink Token Limit Bug (#1284)

* auto-claude: subtask-1-1 - Update THINKING_BUDGET_MAP['ultrathink'] to 64000

- Changed ultrathink token limit from 60000 to 64000 (Claude API maximum)
- Updated comment to reflect it's within API limits (not below)
- Fixes potential API 400 errors when using ultrathink thinking level

* auto-claude: subtask-1-2 - Update frontend THINKING_BUDGET_MAP ultrathink to 64000

* auto-claude: subtask-1-3 - Update test assertions for ultrathink 64000 token limit

Update test_thinking_level_validation.py to expect 64000 for ultrathink
budget, matching the changes made to phase_config.py and the frontend.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>

apps/backend/phase_config.py

@@ -25,7 +25,7 @@ THINKING_BUDGET_MAP: dict[str, int | None] = {
     "low": 1024,
     "medium": 4096,  # Moderate analysis
     "high": 16384,  # Deep thinking for QA review
-    "ultrathink": 60000,  # Maximum reasoning depth (must be < Opus 4.5's 64000 limit)
+    "ultrathink": 64000,  # Maximum reasoning depth (within Claude API's 64000 token limit)
 }
 
 # Spec runner phase-specific thinking levels

apps/frontend/src/shared/constants/models.ts

@@ -28,7 +28,7 @@ export const THINKING_BUDGET_MAP: Record<string, number | null> = {
   low: 1024,
   medium: 4096,
   high: 16384,
-  ultrathink: 60000 // Maximum reasoning depth (must be < Opus 4.5's 64000 limit)
+  ultrathink: 64000 // Maximum reasoning depth (Claude API hard limit)
 } as const;
 
 // ============================================

tests/test_thinking_level_validation.py

@@ -32,8 +32,8 @@ class TestThinkingLevelValidation:
         assert get_thinking_budget("none") is None
 
     def test_ultrathink_max_budget(self):
-        """Test that 'ultrathink' returns maximum budget."""
-        assert get_thinking_budget("ultrathink") == 60000
+        """Test that 'ultrathink' returns maximum budget (Claude API's 64000 token limit)."""
+        assert get_thinking_budget("ultrathink") == 64000
 
     def test_invalid_level_logs_warning(self, caplog):
         """Test that invalid thinking level logs a warning."""
@@ -89,4 +89,4 @@ class TestThinkingLevelValidation:
         assert get_thinking_budget("low") == 1024
         assert get_thinking_budget("medium") == 4096
         assert get_thinking_budget("high") == 16384
-        assert get_thinking_budget("ultrathink") == 60000
+        assert get_thinking_budget("ultrathink") == 64000