main: 增强 Agent Run 调度可靠性与幂等性

- 默认切换 AgentProvider 为 HttpAgentProvider，增强网络请求的容错和重试机制
- 优化 Run 逻辑，支持多场景去重与并发保护
- 添加 Redis 发布失败的日志记录以提升问题排查效率
- 扩展 OpenAPI 规范，新增 Error 和 Run 状态相关模型
- 增强测试覆盖，验证调度策略和重复请求的幂等性
- 增加数据库索引以优化查询性能
- 更新所有相关文档和配置文件

app/Jobs/AgentRunJob.php

@@ -2,6 +2,9 @@
 
 namespace App\Jobs;
 
+use App\Models\Message;
+use App\Services\Agent\ProviderException;
+use App\Services\CancelChecker;
 use App\Services\OutputSink;
 use App\Services\RunLoop;
 use Illuminate\Bus\Queueable;
@@ -14,22 +17,61 @@ class AgentRunJob implements ShouldQueue
 {
     use Dispatchable, InteractsWithQueue, Queueable, SerializesModels;
 
+    public int $tries;
+
+    public int $timeout;
+
+    public int $backoff;
+
     public function __construct(public string $sessionId, public string $runId)
     {
+        $this->tries = (int) config('agent.job.tries', 1);
+        $this->timeout = (int) config('agent.job.timeout_seconds', 120);
+        $this->backoff = (int) config('agent.job.backoff_seconds', 5);
     }
 
-    public function handle(RunLoop $loop, OutputSink $sink): void
+    public function handle(RunLoop $loop, OutputSink $sink, CancelChecker $cancelChecker): void
     {
         try {
+            logger("Running run {$this->runId} for session {$this->sessionId}");
             $loop->run($this->sessionId, $this->runId);
         } catch (\Throwable $e) {
-            $sink->appendError($this->sessionId, $this->runId, 'run.failed', $e->getMessage());
+            logger("Running error  {$this->runId} for session {$this->sessionId}");
+            logger("error message:",[$e->getMessage(),$e->getTraceAsString()]);
+            $errorCode = $e instanceof ProviderException ? $e->errorCode : 'run.failed';
+            $dedupeKey = $e instanceof ProviderException
+                ? "run:{$this->runId}:error:provider"
+                : "run:{$this->runId}:error:job";
+
+            $sink->appendError($this->sessionId, $this->runId, $errorCode, $e->getMessage(), [], $dedupeKey);
             $sink->appendRunStatus($this->sessionId, $this->runId, 'FAILED', [
                 'error' => $e->getMessage(),
                 'dedupe_key' => "run:{$this->runId}:status:FAILED",
             ]);
 
             throw $e;
+        } finally {
+            $latestStatus = Message::query()
+                ->where('session_id', $this->sessionId)
+                ->where('type', 'run.status')
+                ->whereRaw("payload->>'run_id' = ?", [$this->runId])
+                ->orderByDesc('seq')
+                ->first();
+
+            $status = $latestStatus ? ($latestStatus->payload['status'] ?? null) : null;
+
+            if ($status === 'RUNNING' || ! $status) {
+                if ($cancelChecker->isCanceled($this->sessionId, $this->runId)) {
+                    $sink->appendRunStatus($this->sessionId, $this->runId, 'CANCELED', [
+                        'dedupe_key' => "run:{$this->runId}:status:CANCELED",
+                    ]);
+                } else {
+                    $sink->appendRunStatus($this->sessionId, $this->runId, 'FAILED', [
+                        'error' => 'JOB_ENDED_UNEXPECTEDLY',
+                        'dedupe_key' => "run:{$this->runId}:status:FAILED",
+                    ]);
+                }
+            }
         }
     }
 }