改用Grok联网搜索并保留每轮信号

app/agent/runner.py

@@ -13,7 +13,7 @@ from app.db.repository import AgentRepository
 from app.integrations.github_client import GitHubHotProjectClient
 from app.integrations.news_search_client import NewsSearchClient
 from app.llm.grok_client import GrokIntelligenceClient
-from app.schemas.agent import RunResponse
+from app.schemas.agent import RunResponse, SignalItem
 
 logger = logging.getLogger(__name__)
 
@@ -49,7 +49,12 @@ class AgentRunner:
                 github_projects=github_projects,
                 run_date=current_date,
             )
-            signals = self.repo.save_signals(run=run, items=result.signals)
+            signal_items = self._compose_run_signals(
+                news_candidates=news_candidates,
+                github_projects=github_projects,
+                model_signals=result.signals,
+            )
+            signals = self.repo.save_signals(run=run, items=signal_items)
             report_path = self.writer.write_daily_report(current_date, result.report_markdown)
             self.repo.save_report(
                 run=run,
@@ -64,6 +69,8 @@ class AgentRunner:
                     "topic_count": len(topics),
                     "news_candidate_count": len(news_candidates),
                     "github_project_count": len(github_projects),
+                    "model_signal_count": len(result.signals),
+                    "saved_signal_count": len(signal_items),
                     "model": self.grok.settings.llm_model,
                     "model_response": result.raw_response,
                 },
@@ -81,3 +88,29 @@ class AgentRunner:
             self.db.commit()
             logger.exception("agent_run_failed run_id=%s error=%s", run.id, exc)
             raise
+
+    def _compose_run_signals(
+        self,
+        news_candidates: list[SignalItem],
+        github_projects: list[SignalItem],
+        model_signals: list[SignalItem],
+    ) -> list[SignalItem]:
+        # 入库清单以采集候选为准，模型同链接结果只增强摘要和重要度，避免日报模型漏选导致信号消失。
+        model_by_url = {item.source_url: item for item in model_signals}
+        composed = []
+        for candidate in [*news_candidates, *github_projects]:
+            model_item = model_by_url.get(candidate.source_url)
+            if model_item is None:
+                composed.append(candidate)
+                continue
+            composed.append(
+                candidate.model_copy(
+                    update={
+                        "topic": model_item.topic,
+                        "summary": model_item.summary,
+                        "importance": model_item.importance,
+                        "entities": model_item.entities or candidate.entities,
+                    }
+                )
+            )
+        return composed