Fix LLM failure visibility and abort handling

- Update LLM status in real-time as providers fail (not just at end)
- Log individual LLM provider failures to activity log
- Mark "All LLM providers failed" as abort condition (not retryable error)
- Move LLM status update before error check so frontend sees failures
- Show "Aborted" button instead of "Failed-Retry" for LLM failures

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

src/nodes/analyzer.py

@@ -266,15 +266,27 @@ Remember: Every bullet point must reference actual data provided above. Do not i
     response, provider, error, providers_failed = llm.query(prompt, temperature=0)
     elapsed = time.time() - start_time
 
-    # Log failed providers
+    # Log failed providers and update LLM status in real-time
     for pf in providers_failed:
         _add_activity_log(workflow_id, progress_store, "analyzer", f"LLM {pf['name']} failed: {pf['error']}")
+        # Update LLM status in real-time for frontend
+        if workflow_id and progress_store and workflow_id in progress_store:
+            llm_status = progress_store[workflow_id].get("llm_status", {})
+            if pf["name"] in llm_status:
+                llm_status[pf["name"]] = "failed"
 
     # Track failed providers in state for frontend
     if "llm_providers_failed" not in state:
         state["llm_providers_failed"] = []
     state["llm_providers_failed"].extend([pf["name"] for pf in providers_failed])
 
+    # Update successful provider status
+    if provider and workflow_id and progress_store and workflow_id in progress_store:
+        llm_status = progress_store[workflow_id].get("llm_status", {})
+        provider_name = provider.split(":")[0]
+        if provider_name in llm_status:
+            llm_status[provider_name] = "completed"
+
     if error:
         state["draft_report"] = f"Error generating analysis: {error}"
         state["provider_used"] = None

src/services/workflow_store.py

@@ -150,6 +150,7 @@ def run_workflow_background(workflow_id: str, company_name: str, ticker: str, st
                 add_activity_log(workflow_id, source, f"MCP server failed")
 
         # Update LLM status based on failed providers and used provider
+        # IMPORTANT: Do this BEFORE checking for errors so frontend sees failures
         llm_providers_failed = result.get("llm_providers_failed", [])
         provider_used = result.get("provider_used", "")
         llm_status = WORKFLOWS[workflow_id]["llm_status"]
@@ -158,6 +159,7 @@ def run_workflow_background(workflow_id: str, company_name: str, ticker: str, st
         for provider in llm_providers_failed:
             if provider in llm_status:
                 llm_status[provider] = "failed"
+                add_activity_log(workflow_id, "llm", f"{provider.capitalize()} provider failed")
 
         # Mark the used provider as completed
         if provider_used:
@@ -165,6 +167,18 @@ def run_workflow_background(workflow_id: str, company_name: str, ticker: str, st
             if provider_name in llm_status:
                 llm_status[provider_name] = "completed"
 
+        # Check if workflow ended with an error (LLM failures etc)
+        # Do this BEFORE parsing SWOT so we properly abort on errors
+        if result.get("error"):
+            error_msg = result.get("error")
+            add_activity_log(workflow_id, "workflow", f"Workflow failed: {error_msg}")
+            WORKFLOWS[workflow_id].update({
+                "status": "aborted",
+                "error": error_msg,
+                "current_step": "aborted"
+            })
+            return
+
         # Parse SWOT from draft report
         swot_data = parse_swot_text(result.get("draft_report", ""))
 
@@ -186,17 +200,6 @@ def run_workflow_background(workflow_id: str, company_name: str, ticker: str, st
         except Exception as e:
             logger.warning(f"Could not merge MCP SWOT data: {e}")
 
-        # Check if workflow ended with an error (LLM failures etc)
-        if result.get("error"):
-            error_msg = result.get("error")
-            add_activity_log(workflow_id, "workflow", f"Workflow failed: {error_msg}")
-            WORKFLOWS[workflow_id].update({
-                "status": "aborted",
-                "error": error_msg,
-                "current_step": "aborted"
-            })
-            return
-
         # Parse raw_data for MCP display
         raw_data_parsed = {}
         try:
@@ -238,11 +241,12 @@ def run_workflow_background(workflow_id: str, company_name: str, ticker: str, st
     except Exception as e:
         error_msg = str(e)
         # Determine if this is an abort (critical) or error (retryable)
-        # Aborts: Core MCP failures, insufficient data
+        # Aborts: Core MCP failures, insufficient data, LLM failures
         is_abort = any(phrase in error_msg for phrase in [
             "Insufficient core data",
             "All MCP servers failed",
-            "Need at least 2 of"
+            "Need at least 2 of",
+            "All LLM providers failed"
         ])
 
         WORKFLOWS[workflow_id].update({