Add dataset bench and validation agents

backend/app/agents/evaluation_agent.py

@@ -0,0 +1,61 @@
+from __future__ import annotations
+
+from pathlib import Path
+
+from ..catalog import get_catalog
+from ..config import settings
+
+
+REQUIRED_TASKS = {
+    "dataset.upload": "covered_by_api",
+    "dataset.video_frames": "job",
+    "segmodel.train": "job",
+    "segmodel.predict": "job",
+    "yolo.heatmap": "job",
+    "mmseg.flops_fps": "job",
+    "analysis.all": "job",
+}
+
+
+def evaluate_project() -> dict:
+    """Return product/implementation suggestions for the current web platform."""
+    frontend = settings.project_root / "frontend" / "src" / "main.tsx"
+    backend = settings.project_root / "backend" / "app" / "main.py"
+    readme = settings.project_root / "README.md"
+    catalog = get_catalog()
+    checks = []
+    suggestions = []
+
+    frontend_text = frontend.read_text(encoding="utf-8") if frontend.exists() else ""
+    backend_text = backend.read_text(encoding="utf-8") if backend.exists() else ""
+    readme_text = readme.read_text(encoding="utf-8") if readme.exists() else ""
+
+    expectations = {
+        "left_nav_dataset": "数据集" in frontend_text and "#datasets" in frontend_text,
+        "upload_ui": "uploadDatasetFiles" in frontend_text and "labels" in frontend_text and "masks" in frontend_text,
+        "loss_result_ui": "loss" in frontend_text.lower() and "heatmap" in frontend_text.lower(),
+        "dataset_api": "/api/datasets" in backend_text and "api_upload_dataset_files" in backend_text,
+        "no_weight_to_gitea": "Do not push" in readme_text and "check_no_weight_git" in readme_text,
+        "all_core_tasks": all(task in catalog["task_types"] for task in REQUIRED_TASKS if REQUIRED_TASKS[task] == "job"),
+    }
+
+    for name, passed in expectations.items():
+        checks.append({"name": name, "passed": bool(passed)})
+        if not passed:
+            suggestions.append(f"Improve missing capability: {name}")
+
+    if len(catalog["mmseg_algorithms"]) < 31:
+        suggestions.append("MMSeg algorithm catalog should expose all 31 algorithm generators.")
+    if len(catalog["segmodel_architectures"]) < 12:
+        suggestions.append("SegModel catalog should expose all 12 supported architectures.")
+    if not suggestions:
+        suggestions.append("Current platform covers the requested control-plane features; next focus is real dataset/training acceptance tests.")
+
+    score = sum(1 for item in checks if item["passed"]) / max(len(checks), 1)
+    return {
+        "agent": "evaluation_suggestion_agent",
+        "score": round(score, 3),
+        "checks": checks,
+        "suggestions": suggestions,
+    }
+