Expose evaluation agents in the web console

README.md

@@ -155,6 +155,8 @@ Use `GET /api/capabilities` to inspect the grouped full-function readiness
 matrix used by the web dashboard and agents.
 Use `GET /api/results/curves` to inspect parsed training curves discovered
 from YOLO, SegModel, MMSeg, visual-tool, and analysis output directories.
+Use `GET /api/agents/evaluate` and `GET /api/agents/validate` to surface the
+same evaluation and validation feedback shown in the web dashboard Agent panel.
 
 ## Agents
 
@@ -171,3 +173,8 @@ backend/frontend endpoints when the services are running. With live validation
 enabled it also runs the lightweight acceptance smoke above. By default it
 also runs the deep training acceptance; set `SEG_VALIDATE_DEEP=0` when a quick
 non-training validation pass is needed.
+
+The web dashboard calls validation in light mode by default:
+`/api/agents/validate?run_build=false&run_acceptance=false&run_deep=false`.
+Pass `run_acceptance=true` or `run_deep=true` only when you explicitly want the
+agent to launch the heavier runtime acceptance checks from the browser/API.

backend/app/agents/evaluation_agent.py

@@ -56,6 +56,8 @@ def evaluate_project() -> dict:
         "deep_acceptance_api": "/api/acceptance/deep" in backend_text,
         "deep_acceptance_ui": "runDeepAcceptance" in frontend_text and "深度训练" in frontend_text,
         "deep_yolo_heatmap_validation": "yolo_tiny_heatmap_generation" in acceptance_text,
+        "agent_api": "/api/agents/evaluate" in backend_text and "/api/agents/validate" in backend_text,
+        "agent_panel_ui": "runAgentValidation" in frontend_text and "评价建议" in frontend_text and "Validation Agent" in frontend_text,
         "coverage_api": "/api/coverage" in backend_text and coverage["task_build_passed"],
         "visual_tools": "visual.yolo11_heatmap_v2" in catalog["task_types"] and "visual.fps" in catalog["task_types"],
         "yolo_custom_train": "yolo.train_custom" in catalog["task_types"],

backend/app/agents/validation_agent.py

@@ -42,7 +42,7 @@ def _fetch(url: str, timeout: int = 5) -> dict:
         return {"url": url, "error": str(exc), "passed": False}
 
 
-def validate_project(run_build: bool = False) -> dict:
+def validate_project(run_build: bool = False, run_acceptance: bool | None = None, run_deep: bool | None = None) -> dict:
     """Validate current runtime readiness without launching heavy training."""
     checks = []
     catalog = get_catalog()
@@ -140,10 +140,12 @@ def validate_project(run_build: bool = False) -> dict:
         checks.append({"name": "live_coverage_api", "passed": live_coverage["passed"] and '"task_build_passed":true' in live_coverage.get("body", "").replace(" ", ""), "detail": live_coverage})
         checks.append({"name": "live_training_curves_api", "passed": live_curves["passed"] and live_curves.get("body", "").lstrip().startswith("["), "detail": live_curves})
         checks.append({"name": "live_frontend_index", "passed": frontend["passed"] and "Seg Data Server" in frontend.get("body", ""), "detail": frontend})
-        if os.getenv("SEG_VALIDATE_ACCEPTANCE", "1") == "1":
+        acceptance_enabled = run_acceptance if run_acceptance is not None else os.getenv("SEG_VALIDATE_ACCEPTANCE", "1") == "1"
+        deep_enabled = run_deep if run_deep is not None else os.getenv("SEG_VALIDATE_DEEP", "1") == "1"
+        if acceptance_enabled:
             acceptance = run_live_acceptance(backend_url)
             checks.append({"name": "live_acceptance_smoke", "passed": acceptance["passed"], "detail": acceptance})
-        if os.getenv("SEG_VALIDATE_DEEP", "1") == "1":
+        if deep_enabled:
             deep_acceptance = run_deep_acceptance()
             checks.append({"name": "deep_training_acceptance", "passed": deep_acceptance["passed"], "detail": deep_acceptance})
 

backend/app/main.py

@@ -269,5 +269,5 @@ def api_agent_evaluate() -> dict:
 
 
 @app.get("/api/agents/validate")
-def api_agent_validate(run_build: bool = False) -> dict:
-    return validate_project(run_build=run_build)
+def api_agent_validate(run_build: bool = False, run_acceptance: bool = False, run_deep: bool = False) -> dict:
+    return validate_project(run_build=run_build, run_acceptance=run_acceptance, run_deep=run_deep)

frontend/src/main.tsx

@@ -3,6 +3,7 @@ import { createRoot } from "react-dom/client";
 import {
   Activity,
   BarChart3,
+  Bot,
   Boxes,
   ClipboardCheck,
   Cpu,
@@ -231,6 +232,26 @@ type CapabilityPayload = {
   domains: CapabilityDomain[];
 };
 
+type AgentCheck = {
+  name: string;
+  passed: boolean;
+  detail?: unknown;
+  details?: unknown;
+};
+
+type EvaluationAgentPayload = {
+  agent: string;
+  score: number;
+  checks: AgentCheck[];
+  suggestions: string[];
+};
+
+type ValidationAgentPayload = {
+  agent: string;
+  passed: boolean;
+  checks: AgentCheck[];
+};
+
 async function api<T>(path: string, init?: RequestInit): Promise<T> {
   const res = await fetch(`${API_BASE}${path}`, {
     headers: { "Content-Type": "application/json" },
@@ -305,11 +326,12 @@ function useData() {
   const [deepAcceptance, setDeepAcceptance] = useState<DeepAcceptancePayload | null>(null);
   const [runtimeReadiness, setRuntimeReadiness] = useState<RuntimeReadinessPayload | null>(null);
   const [capabilities, setCapabilities] = useState<CapabilityPayload | null>(null);
+  const [agentEvaluation, setAgentEvaluation] = useState<EvaluationAgentPayload | null>(null);
   const [error, setError] = useState<string>("");
 
   async function refresh() {
     try {
-      const [catalogNext, gpusNext, readinessNext, capabilitiesNext, jobsNext, resultsNext, curvesNext, datasetsNext, coverageNext, acceptanceNext, deepAcceptanceNext] = await Promise.all([
+      const [catalogNext, gpusNext, readinessNext, capabilitiesNext, jobsNext, resultsNext, curvesNext, datasetsNext, coverageNext, acceptanceNext, deepAcceptanceNext, agentEvaluationNext] = await Promise.all([
         api<Catalog>("/api/catalog"),
         api<GpuPayload>("/api/system/gpus"),
         api<RuntimeReadinessPayload>("/api/system/readiness"),
@@ -320,7 +342,8 @@ function useData() {
         api<UploadedDataset[]>("/api/datasets"),
         api<CoveragePayload>("/api/coverage"),
         api<AcceptancePayload>("/api/acceptance/latest"),
-        api<DeepAcceptancePayload>("/api/acceptance/deep/latest")
+        api<DeepAcceptancePayload>("/api/acceptance/deep/latest"),
+        api<EvaluationAgentPayload>("/api/agents/evaluate")
       ]);
       setCatalog(catalogNext);
       setGpus(gpusNext);
@@ -345,6 +368,7 @@ function useData() {
       setCoverage(coverageNext);
       setAcceptance(acceptanceNext);
       setDeepAcceptance(deepAcceptanceNext);
+      setAgentEvaluation(agentEvaluationNext);
       setError("");
     } catch (err) {
       setError(String(err));
@@ -357,7 +381,7 @@ function useData() {
     return () => window.clearInterval(timer);
   }, []);
 
-  return { catalog, gpus, runtimeReadiness, capabilities, jobs, results, curves, datasets, datasetValidations, coverage, acceptance, deepAcceptance, error, refresh };
+  return { catalog, gpus, runtimeReadiness, capabilities, agentEvaluation, jobs, results, curves, datasets, datasetValidations, coverage, acceptance, deepAcceptance, error, refresh };
 }
 
 function StatusPill({ status }: { status: string }) {
@@ -380,7 +404,7 @@ function JobProgressBar({ progress }: { progress?: JobProgress }) {
 }
 
 function App() {
-  const { catalog, gpus, runtimeReadiness, capabilities, jobs, results, curves, datasets, datasetValidations, coverage, acceptance, deepAcceptance, error, refresh } = useData();
+  const { catalog, gpus, runtimeReadiness, capabilities, agentEvaluation, jobs, results, curves, datasets, datasetValidations, coverage, acceptance, deepAcceptance, error, refresh } = useData();
   const [taskType, setTaskType] = useState("mock.echo");
   const [params, setParams] = useState(JSON.stringify(defaultParams["mock.echo"], null, 2));
   const [selectedJob, setSelectedJob] = useState<Job | null>(null);
@@ -392,6 +416,8 @@ function App() {
   const [selectedCurvePath, setSelectedCurvePath] = useState("");
   const [uploadKind, setUploadKind] = useState<"images" | "labels" | "masks">("images");
   const [uploadFiles, setUploadFiles] = useState<FileList | null>(null);
+  const [agentValidation, setAgentValidation] = useState<ValidationAgentPayload | null>(null);
+  const [agentBusy, setAgentBusy] = useState(false);
 
   const runningCount = jobs.filter((job) => job.status === "running").length;
   const successCount = jobs.filter((job) => job.status === "success").length;
@@ -491,6 +517,16 @@ function App() {
     }
   }
 
+  async function runAgentValidation() {
+    setAgentBusy(true);
+    try {
+      const result = await api<ValidationAgentPayload>("/api/agents/validate?run_build=false&run_acceptance=false&run_deep=false");
+      setAgentValidation(result);
+    } finally {
+      setAgentBusy(false);
+    }
+  }
+
   async function createDataset() {
     setBusy(true);
     try {
@@ -575,6 +611,7 @@ function App() {
           <a href="#gpu"><Cpu size={18} />GPU</a>
           <a href="#runtime"><ShieldCheck size={18} />环境</a>
           <a href="#coverage"><ClipboardCheck size={18} />覆盖</a>
+          <a href="#agents"><Bot size={18} />Agent</a>
           <a href="#weights"><HardDrive size={18} />权重</a>
           <a href="#results"><BarChart3 size={18} />结果</a>
         </nav>
@@ -892,6 +929,45 @@ function App() {
           </div>
         </section>
 
+        <section className="grid two" id="agents">
+          <div className="panel agentPanel">
+            <div className="panelHead">
+              <div>
+                <p className="eyebrow">Evaluation Agent</p>
+                <h2>评价建议</h2>
+              </div>
+              <StatusPill status={(agentEvaluation?.score ?? 0) >= 1 ? "success" : "queued"} />
+            </div>
+            <div className="agentScore">
+              <strong>{Math.round((agentEvaluation?.score ?? 0) * 100)}%</strong>
+              <span>{agentEvaluation?.checks.filter((item) => item.passed).length ?? 0}/{agentEvaluation?.checks.length ?? 0} checks passed</span>
+            </div>
+            <div className="suggestionList">
+              {(agentEvaluation?.suggestions ?? ["等待评价 agent 返回建议。"]).slice(0, 6).map((item, index) => (
+                <div key={`${index}-${item}`}>{item}</div>
+              ))}
+            </div>
+            <AgentCheckList checks={agentEvaluation?.checks ?? []} limit={14} />
+          </div>
+
+          <div className="panel agentPanel">
+            <div className="panelHead">
+              <div>
+                <p className="eyebrow">Validation Agent</p>
+                <h2>运行验证</h2>
+              </div>
+              <button className="primary" disabled={agentBusy} onClick={runAgentValidation}>
+                <ClipboardCheck size={17} />验证
+              </button>
+            </div>
+            <div className="agentScore">
+              <strong>{agentValidation ? (agentValidation.passed ? "OK" : "Check") : "Ready"}</strong>
+              <span>{agentValidation ? `${agentValidation.checks.filter((item) => item.passed).length}/${agentValidation.checks.length} checks passed` : "轻量验证不会触发深度训练"}</span>
+            </div>
+            <AgentCheckList checks={agentValidation?.checks ?? []} limit={18} />
+          </div>
+        </section>
+
         <section className="grid four">
           <div className="panel" id="gpu">
             <div className="panelHead">
@@ -1047,6 +1123,22 @@ function App() {
   );
 }
 
+function AgentCheckList({ checks, limit }: { checks: AgentCheck[]; limit: number }) {
+  if (!checks.length) {
+    return <p className="muted">尚未运行验证。</p>;
+  }
+  return (
+    <div className="agentCheckList">
+      {checks.slice(0, limit).map((check) => (
+        <div key={check.name} className={check.passed ? "ok" : "bad"} title={check.name}>
+          <span>{check.name}</span>
+          <small>{check.passed ? "passed" : "needs attention"}</small>
+        </div>
+      ))}
+    </div>
+  );
+}
+
 function ResultPreview({ results }: { results: ResultItem[] }) {
   if (!results.length) {
     return <p className="muted">暂无结果</p>;

frontend/src/styles.css

@@ -610,6 +610,82 @@ textarea {
   white-space: nowrap;
 }
 
+.agentPanel {
+  min-height: 360px;
+}
+
+.agentScore {
+  display: grid;
+  grid-template-columns: auto minmax(0, 1fr);
+  align-items: end;
+  gap: 12px;
+  margin-bottom: 12px;
+}
+
+.agentScore strong {
+  font-size: 38px;
+  line-height: 1;
+}
+
+.agentScore span {
+  min-width: 0;
+  color: var(--muted);
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.suggestionList,
+.agentCheckList {
+  display: grid;
+  gap: 8px;
+}
+
+.suggestionList {
+  margin-bottom: 12px;
+}
+
+.suggestionList div,
+.agentCheckList div {
+  min-width: 0;
+  padding: 9px;
+  border-radius: 6px;
+  border: 1px solid var(--line);
+  background: #101310;
+}
+
+.suggestionList div {
+  color: var(--ink);
+  line-height: 1.45;
+}
+
+.agentCheckList {
+  grid-template-columns: repeat(2, minmax(0, 1fr));
+  max-height: 310px;
+  overflow: auto;
+}
+
+.agentCheckList div.ok {
+  border-color: rgba(157, 226, 111, 0.32);
+}
+
+.agentCheckList div.bad {
+  border-color: rgba(240, 113, 103, 0.55);
+}
+
+.agentCheckList span,
+.agentCheckList small {
+  display: block;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
+.agentCheckList small {
+  color: var(--muted);
+  margin-top: 2px;
+}
+
 .datasetCard {
   width: 100%;
   display: block;
@@ -1126,6 +1202,7 @@ meter {
   .opGrid,
   .sampleStrip,
   .taskCheckList,
+  .agentCheckList,
   .qualityStats,
   .qualityChecks {
     grid-template-columns: repeat(2, minmax(0, 1fr));