From c1583820ea675e882dc3d167b77ee7c3f74f380c Mon Sep 17 00:00:00 2001
From: jlightner <jlightner@users.noreply.github.com>
Date: Tue, 31 Mar 2026 17:27:40 +0000
Subject: [PATCH] feat: Add context labels to multi-call pipeline stages

Stage 3 (extraction) LLM calls now show the topic group label (e.g.,
'Sound Design Basics') and Stage 5 (synthesis) calls show the category
name. Displayed as a cyan italic label in the event row between the
event type badge and model name. Helps admins understand why there are
multiple LLM calls per stage.
---
 backend/pipeline/stages.py           |  6 ++++--
 frontend/src/App.css                 | 10 ++++++++++
 frontend/src/pages/AdminPipeline.tsx |  5 +++++
 3 files changed, 19 insertions(+), 2 deletions(-)

diff --git a/backend/pipeline/stages.py b/backend/pipeline/stages.py
index ec99333..c8ff245 100644
--- a/backend/pipeline/stages.py
+++ b/backend/pipeline/stages.py
@@ -135,6 +135,7 @@ def _make_llm_callback(
     system_prompt: str | None = None,
     user_prompt: str | None = None,
     run_id: str | None = None,
+    context_label: str | None = None,
 ):
     """Create an on_complete callback for LLMClient that emits llm_call events.
 
@@ -162,6 +163,7 @@ def _make_llm_callback(
                 "content_length": len(content) if content else 0,
                 "finish_reason": finish_reason,
                 "is_fallback": is_fallback,
+                **({"context": context_label} if context_label else {}),
             },
             system_prompt_text=system_prompt if debug else None,
             user_prompt_text=user_prompt if debug else None,
@@ -427,7 +429,7 @@ def stage3_extraction(self, video_id: str, run_id: str | None = None) -> str:
             )
 
             max_tokens = estimate_max_tokens(system_prompt, user_prompt, stage="stage3_extraction", hard_limit=hard_limit)
-            raw = llm.complete(system_prompt, user_prompt, response_model=ExtractionResult, on_complete=_make_llm_callback(video_id, "stage3_extraction", system_prompt=system_prompt, user_prompt=user_prompt, run_id=run_id),
+            raw = llm.complete(system_prompt, user_prompt, response_model=ExtractionResult, on_complete=_make_llm_callback(video_id, "stage3_extraction", system_prompt=system_prompt, user_prompt=user_prompt, run_id=run_id, context_label=topic_label),
                                modality=modality, model_override=model_override, max_tokens=max_tokens)
             result = _safe_parse_llm_response(raw, ExtractionResult, llm, system_prompt, user_prompt,
                                               modality=modality, model_override=model_override)
@@ -781,7 +783,7 @@ def stage5_synthesis(self, video_id: str, run_id: str | None = None) -> str:
             user_prompt = f"<creator>{creator_name}</creator>\n<moments>\n{moments_text}\n</moments>"
 
             max_tokens = estimate_max_tokens(system_prompt, user_prompt, stage="stage5_synthesis", hard_limit=hard_limit)
-            raw = llm.complete(system_prompt, user_prompt, response_model=SynthesisResult, on_complete=_make_llm_callback(video_id, "stage5_synthesis", system_prompt=system_prompt, user_prompt=user_prompt, run_id=run_id),
+            raw = llm.complete(system_prompt, user_prompt, response_model=SynthesisResult, on_complete=_make_llm_callback(video_id, "stage5_synthesis", system_prompt=system_prompt, user_prompt=user_prompt, run_id=run_id, context_label=category),
                                modality=modality, model_override=model_override, max_tokens=max_tokens)
             result = _safe_parse_llm_response(raw, SynthesisResult, llm, system_prompt, user_prompt,
                                               modality=modality, model_override=model_override)
diff --git a/frontend/src/App.css b/frontend/src/App.css
index bab480d..067fa42 100644
--- a/frontend/src/App.css
+++ b/frontend/src/App.css
@@ -4133,6 +4133,16 @@ a.app-footer__repo:hover {
   font-weight: 500;
 }
 
+.pipeline-event__context {
+  color: var(--color-accent);
+  font-size: 0.75rem;
+  font-style: italic;
+  max-width: 150px;
+  overflow: hidden;
+  text-overflow: ellipsis;
+  white-space: nowrap;
+}
+
 .pipeline-event__model {
   color: var(--color-text-muted);
   font-size: 0.75rem;
diff --git a/frontend/src/pages/AdminPipeline.tsx b/frontend/src/pages/AdminPipeline.tsx
index 0c6b657..66cfa44 100644
--- a/frontend/src/pages/AdminPipeline.tsx
+++ b/frontend/src/pages/AdminPipeline.tsx
@@ -294,6 +294,11 @@ function EventLog({ videoId, status, runId }: { videoId: string; status: string;
               <span className={`pipeline-badge pipeline-badge--event-${evt.event_type}`}>
                 {evt.event_type}
               </span>
+              {typeof evt.payload?.context === "string" && (
+                <span className="pipeline-event__context" title="Processing context">
+                  {evt.payload.context}
+                </span>
+              )}
               {evt.model && <span className="pipeline-event__model">{evt.model}</span>}
               {evt.total_tokens != null && evt.total_tokens > 0 && (
                 <span className="pipeline-event__tokens" title={`prompt: ${evt.prompt_tokens ?? 0} / completion: ${evt.completion_tokens ?? 0}`}>