feat: add tool-call fallback and resilient batch exports

apps/admin_console/routers/system.py

@@ -517,9 +517,12 @@ async def get_model_config_and_env():
 
 @router.get("/server-status")
 async def get_server_runtime_status():
-    """Retrieve runtime status, PID, port, and uptime of the Artemis server."""
+    """Return this server's status without scanning listening ports."""
     import os
-    from artemis.runtime.server_lifecycle import get_server_status
+    import time
+
+    from artemis.runtime.process_probe import pid_is_alive
+    from artemis.runtime.server_lifecycle import read_server_info
 
     try:
         from apps.admin_console.core.state import state
@@ -527,18 +530,34 @@ async def get_server_runtime_status():
         from admin_console.core.state import state
 
     port = getattr(state, "port", 8000)
-    status = get_server_status(port=port)
+    current_pid = os.getpid()
+    pids = {current_pid}
+    started_at = None
+    info = read_server_info()
+    if info and info.get("port") == port:
+        saved_pid = info.get("pid")
+        if isinstance(saved_pid, int) and saved_pid != current_pid and pid_is_alive(saved_pid):
+            pids.add(saved_pid)
+        if saved_pid == current_pid and isinstance(info.get("started_at"), (int, float)):
+            started_at = float(info["started_at"])
+    if started_at is None:
+        try:
+            import psutil
+
+            started_at = psutil.Process(current_pid).create_time()
+        except Exception:
+            started_at = None
     # Explicit DTO: the raw metadata file additionally holds the lifecycle
     # token, cmdline, and filesystem paths, none of which belong on the wire.
     return {
-        "running": status["running"],
-        "port": status["port"],
-        "pids": status["pids"],
-        "active_pid": status["active_pid"],
-        "uptime_seconds": status["uptime_seconds"],
-        "url": status["url"],
-        "admin_url": status["admin_url"],
-        "current_pid": os.getpid(),
+        "running": True,
+        "port": port,
+        "pids": sorted(pids),
+        "active_pid": current_pid,
+        "uptime_seconds": max(0.0, time.time() - started_at) if started_at else None,
+        "url": f"http://localhost:{port}",
+        "admin_url": f"http://localhost:{port}/admin",
+        "current_pid": current_pid,
     }
 
 

apps/admin_console/services/model_settings_service.py

@@ -23,6 +23,13 @@ class ModelSettingsUpdate(BaseModel):
     timeout_seconds: float = Field(default=300, gt=0, le=3600)
     max_retries: int = Field(default=5, ge=0, le=10)
     retry_interval_seconds: float = Field(default=10, ge=0, le=300)
+    rate_limit_retry_seconds: float | None = Field(default=None, ge=0, le=3600)
+    tool_fallback_model: str | None = Field(default=None, min_length=1, max_length=200)
+    tool_fallback_reasoning_effort: (
+        Literal["default", "none", "minimal", "low", "medium", "high", "xhigh", "max"] | None
+    ) = None
+    tool_fallback_max_retries: int | None = Field(default=None, ge=0, le=10)
+    tool_fallback_no_progress_attempts: int | None = Field(default=None, ge=1, le=10)
     reasoning_effort: Literal[
         "default", "none", "minimal", "low", "medium", "high", "xhigh", "max"
     ] = "high"
@@ -45,9 +52,11 @@ class ModelSettingsUpdate(BaseModel):
             raise ValueError("请输入接口根地址，例如 https://api-n-cd.playones.com/v1")
         return value
 
-    @field_validator("model")
+    @field_validator("model", "tool_fallback_model")
     @classmethod
-    def validate_model(cls, value: str) -> str:
+    def validate_model(cls, value: str | None) -> str | None:
+        if value is None:
+            return None
         value = value.strip()
         if not value or any(c in value for c in "\r\n\x00"):
             raise ValueError("模型名称不能为空或包含换行")
@@ -64,6 +73,11 @@ def get_model_settings() -> dict:
         "timeout_seconds": settings.OPENAI_TIMEOUT_SECONDS,
         "max_retries": settings.OPENAI_MAX_RETRIES,
         "retry_interval_seconds": settings.OPENAI_RETRY_INTERVAL_SECONDS,
+        "rate_limit_retry_seconds": settings.OPENAI_RATE_LIMIT_RETRY_SECONDS,
+        "tool_fallback_model": settings.OPENAI_TOOL_FALLBACK_MODEL,
+        "tool_fallback_reasoning_effort": settings.OPENAI_TOOL_FALLBACK_REASONING_EFFORT,
+        "tool_fallback_max_retries": settings.OPENAI_TOOL_FALLBACK_MAX_RETRIES,
+        "tool_fallback_no_progress_attempts": settings.ARTEMIS_TOOL_FALLBACK_MAX_NO_PROGRESS_ATTEMPTS,
         "reasoning_effort": settings.OPENAI_REASONING_EFFORT,
         "model_presets": [
             "gpt-5.6-luna",
@@ -86,6 +100,20 @@ def save_model_settings(update: ModelSettingsUpdate) -> dict:
         "OPENAI_RETRY_INTERVAL_SECONDS": str(update.retry_interval_seconds),
         "OPENAI_REASONING_EFFORT": update.reasoning_effort,
     }
+    optional_settings = {
+        "rate_limit_retry_seconds": ("OPENAI_RATE_LIMIT_RETRY_SECONDS", float),
+        "tool_fallback_model": ("OPENAI_TOOL_FALLBACK_MODEL", str),
+        "tool_fallback_reasoning_effort": ("OPENAI_TOOL_FALLBACK_REASONING_EFFORT", str),
+        "tool_fallback_max_retries": ("OPENAI_TOOL_FALLBACK_MAX_RETRIES", int),
+        "tool_fallback_no_progress_attempts": (
+            "ARTEMIS_TOOL_FALLBACK_MAX_NO_PROGRESS_ATTEMPTS",
+            int,
+        ),
+    }
+    for field_name, (env_name, expected_type) in optional_settings.items():
+        value = getattr(update, field_name)
+        if value is not None:
+            values[env_name] = str(value)
     if update.api_key and update.api_key.get_secret_value().strip():
         values["OPENAI_API_KEY"] = update.api_key.get_secret_value().strip()
     persist_environment(values)
@@ -95,6 +123,10 @@ def save_model_settings(update: ModelSettingsUpdate) -> dict:
     settings.OPENAI_MAX_RETRIES = update.max_retries
     settings.OPENAI_RETRY_INTERVAL_SECONDS = update.retry_interval_seconds
     settings.OPENAI_REASONING_EFFORT = update.reasoning_effort
+    for field_name, (env_name, expected_type) in optional_settings.items():
+        value = getattr(update, field_name)
+        if value is not None:
+            setattr(settings, env_name, expected_type(value))
     if "OPENAI_API_KEY" in values:
         settings.OPENAI_API_KEY = SecretStr(values["OPENAI_API_KEY"])
     from artemis.llm.router import ModelFactory

apps/admin_console/services/report_export.py

@@ -734,9 +734,12 @@ def export_batch(batch_id: str) -> Path:
             for path in outputs:
                 archive.write(path, path.name)
             for number, report in enumerate(records, 1):
+                prefix = f"任务报告/{number:04d}_{archive_case_name(report)}"
                 for path in export_session(report["id"], formats):
-                    archive.write(
-                        path, f"任务报告/{number:04d}_{archive_case_name(report)}/{path.name}"
-                    )
+                    archive.write(path, f"{prefix}/{path.name}")
+                for source, name in session_screenshots(report):
+                    archive.write(source, f"{prefix}/步骤截图/{name}")
+                for source, name in session_videos(report):
+                    archive.write(source, f"{prefix}/任务录屏/{name}")
         temporary.replace(target)
         return target

apps/showcase_ui/src/app/components/model-settings/model-settings.component.spec.ts

@@ -8,6 +8,8 @@ describe('持久化模型与功能设置', () => {
   let http: HttpTestingController;
   const config = { base_url: 'https://example.test/v1', model: 'mimo-v2.6-flash',
     api_key_configured: true, timeout_seconds: 300, max_retries: 5, retry_interval_seconds: 10, reasoning_effort: 'high',
+    rate_limit_retry_seconds: 60, tool_fallback_model: 'gpt-6-luna', tool_fallback_reasoning_effort: 'high',
+    tool_fallback_max_retries: 3, tool_fallback_no_progress_attempts: 3,
     model_presets: ['gpt-5.6-luna', 'gpt-6-luna', 'mimo-v2.6-flash', 'mimo-v2.6-pro'], env_path: '/app/.env' };
   beforeEach(() => {
     TestBed.configureTestingModule({ providers: [provideHttpClient(), provideHttpClientTesting()] });
@@ -31,6 +33,9 @@ describe('持久化模型与功能设置', () => {
     expect(request.request.body.timeout_seconds).toBe(300);
     expect(request.request.body.max_retries).toBe(5);
     expect(request.request.body.retry_interval_seconds).toBe(10);
+    expect(request.request.body.rate_limit_retry_seconds).toBe(60);
+    expect(request.request.body.tool_fallback_model).toBe('gpt-6-luna');
+    expect(request.request.body.tool_fallback_no_progress_attempts).toBe(3);
     expect(request.request.body.api_key).toBeUndefined();
     request.flush({ ...config, model: 'another-model' });
     expect(component.busy()).toBeNull();
@@ -88,6 +93,29 @@ describe('持久化模型与功能设置', () => {
     expect(reopened.nativeElement.querySelector('#model-reasoning').value).toBe('max');
   });
 
+  it('高级稳定性设置可编辑并保存', async () => {
+    const fixture = TestBed.createComponent(ModelSettingsComponent);
+    fixture.detectChanges();
+    http.expectOne('/api/system/model-settings').flush(config);
+    fixture.detectChanges();
+    await fixture.whenStable();
+    const root: HTMLElement = fixture.nativeElement;
+    root.querySelector<HTMLDetailsElement>('.recovery-settings')!.open = true;
+    const fallback = root.querySelector<HTMLInputElement>('input[name="toolFallbackModel"]')!;
+    fallback.value = 'custom/fallback';
+    fallback.dispatchEvent(new Event('input'));
+    const wait = root.querySelector<HTMLInputElement>('input[name="rateLimitRetry"]')!;
+    wait.value = '90';
+    wait.dispatchEvent(new Event('input'));
+    fixture.detectChanges();
+    await fixture.whenStable();
+    root.querySelector<HTMLFormElement>('form')!.dispatchEvent(new Event('submit', { cancelable: true }));
+    const request = http.expectOne('/api/system/model-settings');
+    expect(request.request.body.tool_fallback_model).toBe('custom/fallback');
+    expect(request.request.body.rate_limit_retry_seconds).toBe(90);
+    request.flush({ ...config, tool_fallback_model: 'custom/fallback', rate_limit_retry_seconds: 90 });
+  });
+
   it('连接测试使用当前输入而不写入保存接口', () => {
     const fixture = TestBed.createComponent(ModelSettingsComponent);
     fixture.detectChanges();

apps/showcase_ui/src/app/components/model-settings/model-settings.component.ts

@@ -10,6 +10,11 @@ interface ModelSettings {
   timeout_seconds: number;
   max_retries: number;
   retry_interval_seconds: number;
+  rate_limit_retry_seconds: number;
+  tool_fallback_model: string;
+  tool_fallback_reasoning_effort: string;
+  tool_fallback_max_retries: number;
+  tool_fallback_no_progress_attempts: number;
   reasoning_effort: string;
   model_presets: string[];
   env_path: string;
@@ -56,6 +61,30 @@ interface ModelSettings {
           [(ngModel)]="retries" [disabled]="busy() || loading()" /></label>
         <label>重试间隔（秒）<input name="retryInterval" type="number" min="0" max="300" required [(ngModel)]="retryInterval" [disabled]="busy() || loading()" /></label>
       </div>
+      <details class="recovery-settings">
+        <summary>稳定性与兜底模型</summary>
+        <div class="recovery-content">
+          <label>429 等待重试（秒）<input name="rateLimitRetry" type="number" min="0" max="3600" required
+            [(ngModel)]="rateLimitRetry" [disabled]="busy() || loading()" /></label>
+          <label>兜底模型名称<input name="toolFallbackModel" type="text" required autocomplete="off"
+            autocapitalize="off" spellcheck="false" [(ngModel)]="toolFallbackModel"
+            [disabled]="busy() || loading()" /></label>
+          <label>兜底思考强度<select name="toolFallbackReasoning" [(ngModel)]="toolFallbackReasoning"
+            [disabled]="busy() || loading()">
+            <option value="default">由模型默认</option><option value="none">关闭（none）</option>
+            <option value="minimal">最低（minimal）</option><option value="low">低（low）</option>
+            <option value="medium">中（medium）</option><option value="high">高（high）</option>
+            <option value="xhigh">极高（xhigh）</option><option value="max">最高（max）</option>
+          </select></label>
+          <div class="number-fields">
+            <label>兜底请求重试次数<input name="toolFallbackRetries" type="number" min="0" max="10" required
+              [(ngModel)]="toolFallbackRetries" [disabled]="busy() || loading()" /></label>
+            <label>兜底无效调用上限<input name="toolFallbackNoProgress" type="number" min="1" max="10" required
+              [(ngModel)]="toolFallbackNoProgress" [disabled]="busy() || loading()" /></label>
+          </div>
+          <small>首选模型连续 3 次未形成有效工具调用后，本步骤使用兜底模型；下一步恢复首选模型。</small>
+        </div>
+      </details>
       <div class="actions">
         <button type="submit" [disabled]="busy() || loading() || !valid()">{{ busy() === 'save' ? '正在保存…' : '保存配置' }}</button>
         <button type="button" (click)="test()" [disabled]="busy() || loading() || !valid()">
@@ -77,6 +106,10 @@ interface ModelSettings {
     input:focus, select:focus { outline: 2px solid #38bdf8; }
     .number-fields { display: grid; grid-template-columns: 1fr 1fr; gap: 16px; margin-top: 8px; }
     .number-fields input { margin-top: 8px; }
+    .recovery-settings { border: 1px solid #cbd5e1; border-radius: 10px; background: #f8fafc; padding: 10px 14px; }
+    .recovery-settings summary { cursor: pointer; font-weight: 600; color: #334155; }
+    .recovery-content { display: grid; gap: 10px; padding-top: 12px; }
+    .recovery-content label input, .recovery-content label select { margin-top: 8px; }
     .actions { display: flex; gap: 12px; margin-top: 8px; }
     button { padding: 11px 20px; border-radius: 10px; border: 1px solid #38bdf877; background: #0284c7; color: white; cursor: pointer; font: inherit; }
     button:disabled { opacity: .5; cursor: default; } .success { color: #047857; } .error { color: #be123c; }
@@ -93,6 +126,8 @@ export class ModelSettingsComponent implements OnInit {
   saved = output<void>();
   modelPresets: string[] = [];
   baseUrl = ''; model = ''; apiKey = ''; timeout = 300; retries = 5; retryInterval = 10; reasoning = 'high';
+  rateLimitRetry = 60; toolFallbackModel = 'gpt-6-luna'; toolFallbackReasoning = 'high';
+  toolFallbackRetries = 3; toolFallbackNoProgress = 3;
   busy = signal<'save' | 'test' | null>(null);
   loading = signal(true); keyConfigured = signal(false); envPath = signal('');
   message = signal(''); error = signal('');
@@ -105,16 +140,27 @@ export class ModelSettingsComponent implements OnInit {
   }
   valid(): boolean {
     return !!this.baseUrl.trim() && !!this.model.trim() && (!!this.apiKey.trim() || this.keyConfigured())
-      && this.timeout > 0 && this.timeout <= 3600 && Number.isInteger(this.retries) && this.retries >= 0 && this.retries <= 10 && this.retryInterval >= 0 && this.retryInterval <= 300;
+      && this.timeout > 0 && this.timeout <= 3600 && Number.isInteger(this.retries) && this.retries >= 0 && this.retries <= 10 && this.retryInterval >= 0 && this.retryInterval <= 300
+      && this.rateLimitRetry >= 0 && this.rateLimitRetry <= 3600 && !!this.toolFallbackModel.trim()
+      && Number.isInteger(this.toolFallbackRetries) && this.toolFallbackRetries >= 0 && this.toolFallbackRetries <= 10
+      && Number.isInteger(this.toolFallbackNoProgress) && this.toolFallbackNoProgress >= 1 && this.toolFallbackNoProgress <= 10;
   }
   private payload() {
     return { base_url: this.baseUrl.trim(), model: this.model.trim(), api_key: this.apiKey.trim() || undefined,
-      timeout_seconds: this.timeout, max_retries: this.retries, retry_interval_seconds: this.retryInterval, reasoning_effort: this.reasoning };
+      timeout_seconds: this.timeout, max_retries: this.retries, retry_interval_seconds: this.retryInterval, reasoning_effort: this.reasoning,
+      rate_limit_retry_seconds: this.rateLimitRetry, tool_fallback_model: this.toolFallbackModel.trim(),
+      tool_fallback_reasoning_effort: this.toolFallbackReasoning, tool_fallback_max_retries: this.toolFallbackRetries,
+      tool_fallback_no_progress_attempts: this.toolFallbackNoProgress };
   }
   private apply(config: ModelSettings): void {
     this.baseUrl = config.base_url; this.model = config.model; this.timeout = config.timeout_seconds;
     this.modelPresets = config.model_presets;
     this.retries = config.max_retries; this.retryInterval = config.retry_interval_seconds ?? 10; this.reasoning = config.reasoning_effort ?? 'high'; this.keyConfigured.set(config.api_key_configured); this.envPath.set(config.env_path);
+    this.rateLimitRetry = config.rate_limit_retry_seconds ?? 60;
+    this.toolFallbackModel = config.tool_fallback_model ?? 'gpt-6-luna';
+    this.toolFallbackReasoning = config.tool_fallback_reasoning_effort ?? 'high';
+    this.toolFallbackRetries = config.tool_fallback_max_retries ?? 3;
+    this.toolFallbackNoProgress = config.tool_fallback_no_progress_attempts ?? 3;
   }
   save(): void {
     if (this.busy() || !this.valid()) return;