fix: address remaining review comments — broken import, race condition, duplicate logic

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

lib/cli/src/crewai_cli/agent_tui.py

@@ -1035,6 +1035,34 @@ class AgentTUI(App[None]):
             return self._get_or_create_agent(self._agent_names[0])
         return None
 
+    def _format_memory_record(self, i: int, mem: object) -> list[str]:
+        """Format a single memory record into Rich markup lines."""
+        record = getattr(mem, "record", mem)
+        content = getattr(record, "content", "") or str(mem)
+        if len(content) > 150:
+            content = content[:150] + "..."
+
+        meta = getattr(record, "metadata", {}) or {}
+        mem_type = meta.get("type", "raw")
+        importance = getattr(record, "importance", "") or meta.get("importance", "")
+        scope = getattr(record, "scope", "") or meta.get("scope", "")
+        timestamp = getattr(record, "created_at", "")
+
+        type_tag = (
+            f"[bold cyan]{mem_type}[/]"
+            if mem_type == "canonical"
+            else f"[dim]{mem_type}[/]"
+        )
+        importance_tag = f" [yellow]★{importance}[/]" if importance else ""
+        scope_tag = f" [{_DIM}]scope:{scope}[/]" if scope else ""
+        time_tag = f" [{_DIM}]{timestamp}[/]" if timestamp else ""
+
+        return [
+            f"  {i}. {type_tag}{importance_tag}{scope_tag}{time_tag}",
+            f"     {content}",
+            "",
+        ]
+
     def _show_memory_panel(self) -> None:
         """Show recent memories for the focused agent (GAP-92: rich formatting)."""
         agent = self._get_focused_agent()
@@ -1055,31 +1083,7 @@ class AgentTUI(App[None]):
             lines = [f"[bold]Memory Inspector — {agent_name}[/]\n"]
 
             for i, mem in enumerate(memories, 1):
-                record = getattr(mem, "record", mem)
-                content = getattr(record, "content", "") or str(mem)
-                if len(content) > 150:
-                    content = content[:150] + "..."
-
-                meta = getattr(record, "metadata", {}) or {}
-                mem_type = meta.get("type", "raw")
-                importance = getattr(record, "importance", "") or meta.get("importance", "")
-                scope = getattr(record, "scope", "") or meta.get("scope", "")
-                timestamp = getattr(record, "created_at", "")
-
-                type_tag = (
-                    f"[bold cyan]{mem_type}[/]"
-                    if mem_type == "canonical"
-                    else f"[dim]{mem_type}[/]"
-                )
-                importance_tag = f" [yellow]★{importance}[/]" if importance else ""
-                scope_tag = f" [{_DIM}]scope:{scope}[/]" if scope else ""
-                time_tag = f" [{_DIM}]{timestamp}[/]" if timestamp else ""
-
-                lines.append(
-                    f"  {i}. {type_tag}{importance_tag}{scope_tag}{time_tag}"
-                )
-                lines.append(f"     {content}")
-                lines.append("")
+                lines.extend(self._format_memory_record(i, mem))
 
             lines.append(f"[{_DIM}]Use /memory search <query> to filter[/]")
             self._mount_sys("\n".join(lines))
@@ -1106,31 +1110,7 @@ class AgentTUI(App[None]):
             lines = [f"[bold]Memories matching '{query}' — {agent_name}[/]\n"]
 
             for i, mem in enumerate(results, 1):
-                record = getattr(mem, "record", mem)
-                content = getattr(record, "content", "") or str(mem)
-                if len(content) > 150:
-                    content = content[:150] + "..."
-
-                meta = getattr(record, "metadata", {}) or {}
-                mem_type = meta.get("type", "raw")
-                importance = getattr(record, "importance", "") or meta.get("importance", "")
-                scope = getattr(record, "scope", "") or meta.get("scope", "")
-                timestamp = getattr(record, "created_at", "")
-
-                type_tag = (
-                    f"[bold cyan]{mem_type}[/]"
-                    if mem_type == "canonical"
-                    else f"[dim]{mem_type}[/]"
-                )
-                importance_tag = f" [yellow]★{importance}[/]" if importance else ""
-                scope_tag = f" [{_DIM}]scope:{scope}[/]" if scope else ""
-                time_tag = f" [{_DIM}]{timestamp}[/]" if timestamp else ""
-
-                lines.append(
-                    f"  {i}. {type_tag}{importance_tag}{scope_tag}{time_tag}"
-                )
-                lines.append(f"     {content}")
-                lines.append("")
+                lines.extend(self._format_memory_record(i, mem))
 
             lines.append(f"[{_DIM}]Use /memory search <query> to refine[/]")
             self._mount_sys("\n".join(lines))

lib/cli/src/crewai_cli/benchmark.py

@@ -197,10 +197,8 @@ async def _run_model_benchmark(
     emit({"type": "model_start", "model": model, "total_cases": total})
 
     sem = asyncio.Semaphore(_MAX_CASES_PARALLEL)
-    done_count = 0
 
     async def _run_case(i: int, case: BenchmarkCase) -> BenchmarkResult:
-        nonlocal done_count
         async with sem:
             emit({"type": "case_start", "model": model, "case_index": i, "total_cases": total, "input": case.input})
 
@@ -220,8 +218,7 @@ async def _run_model_benchmark(
             try:
                 agent = _load_agent(bench_defn, agents_dir=agents_dir)
             except Exception as e:
-                done_count += 1
-                emit({"type": "case_done", "model": model, "case_index": done_count, "total_cases": total, "passed": False, "score": 0.0, "time_ms": 0, "error": str(e)})
+                emit({"type": "case_done", "model": model, "case_index": i, "total_cases": total, "passed": False, "score": 0.0, "time_ms": 0, "error": str(e)})
                 return BenchmarkResult(
                     case_index=i, input=case.input, expected=case.expected,
                     actual=f"[Agent creation error: {e}]", model=model, passed=False, score=0.0,
@@ -240,8 +237,7 @@ async def _run_model_benchmark(
                 cost = response.cost
             except asyncio.TimeoutError:
                 elapsed_ms = _current_time_ms() - start_ms
-                done_count += 1
-                emit({"type": "case_done", "model": model, "case_index": done_count, "total_cases": total, "passed": False, "score": 0.0, "time_ms": elapsed_ms, "error": "timeout"})
+                emit({"type": "case_done", "model": model, "case_index": i, "total_cases": total, "passed": False, "score": 0.0, "time_ms": elapsed_ms, "error": "timeout"})
                 return BenchmarkResult(
                     case_index=i, input=case.input, expected=case.expected,
                     actual=f"[Timeout after {_CASE_TIMEOUT_SECONDS}s]", model=model, passed=False, score=0.0,
@@ -249,8 +245,7 @@ async def _run_model_benchmark(
                 )
             except Exception as e:
                 elapsed_ms = _current_time_ms() - start_ms
-                done_count += 1
-                emit({"type": "case_done", "model": model, "case_index": done_count, "total_cases": total, "passed": False, "score": 0.0, "time_ms": elapsed_ms, "error": str(e)})
+                emit({"type": "case_done", "model": model, "case_index": i, "total_cases": total, "passed": False, "score": 0.0, "time_ms": elapsed_ms, "error": str(e)})
                 return BenchmarkResult(
                     case_index=i, input=case.input, expected=case.expected,
                     actual=f"[Error: {e}]", model=model, passed=False, score=0.0,
@@ -261,7 +256,7 @@ async def _run_model_benchmark(
             if case.expected is not None:
                 passed, score = _check_expected(case.expected, actual)
             if case.criteria is not None:
-                emit({"type": "judging", "model": model, "case_index": done_count + 1, "total_cases": total})
+                emit({"type": "judging", "model": model, "case_index": i, "total_cases": total})
                 try:
                     criteria_passed, criteria_score = await asyncio.wait_for(
                         _judge_with_llm(case.criteria, case.input, actual, judge_model),
@@ -275,8 +270,7 @@ async def _run_model_benchmark(
                 else:
                     passed, score = criteria_passed, criteria_score
 
-            done_count += 1
-            emit({"type": "case_done", "model": model, "case_index": done_count, "total_cases": total, "passed": passed, "score": score, "time_ms": elapsed_ms})
+            emit({"type": "case_done", "model": model, "case_index": i, "total_cases": total, "passed": passed, "score": score, "time_ms": elapsed_ms})
             return BenchmarkResult(
                 case_index=i, input=case.input, expected=case.expected, actual=actual,
                 model=model, passed=passed, score=score,

lib/cli/src/crewai_cli/cli.py

@@ -229,8 +229,8 @@ def _train_new_agents(agent_files: list, n_iterations: int) -> None:
         click.secho(f"Training {agent_name} ({len(cases)} cases, {n_iterations} iterations)", fg="cyan", bold=True)
 
         try:
-            from crewai.new_agent.definition_parser import load_agent_definition
-            agent = load_agent_definition(str(agent_path))
+            from crewai.new_agent.definition_parser import load_agent_from_definition
+            agent = load_agent_from_definition(str(agent_path))
         except Exception as e:
             click.secho(f"  Error loading agent {agent_name}: {e}", fg="red")
             continue