From ad4e391ca5f7355338a04fece2b1d3815371d358 Mon Sep 17 00:00:00 2001
From: Your Name <joshualitt@google.com>
Date: Fri, 10 Apr 2026 17:51:27 +0000
Subject: [PATCH] refactor(context): inline lightweight summarization logic
 inside rollingSummaryProcessor to break dependency on snapshotGenerator

---
 .../processors/rollingSummaryProcessor.ts     | 44 ++++++++++++++++---
 1 file changed, 38 insertions(+), 6 deletions(-)

diff --git a/packages/core/src/context/processors/rollingSummaryProcessor.ts b/packages/core/src/context/processors/rollingSummaryProcessor.ts
index 8ef10af38c..21f57eb048 100644
--- a/packages/core/src/context/processors/rollingSummaryProcessor.ts
+++ b/packages/core/src/context/processors/rollingSummaryProcessor.ts
@@ -10,8 +10,8 @@ import type {
 } from '../pipeline.js';
 import type { ContextEnvironment } from '../pipeline/environment.js';
 import type { ConcreteNode, RollingSummary } from '../ir/types.js';
-import { SnapshotGenerator } from '../utils/snapshotGenerator.js';
 import { debugLogger } from '../../utils/debugLogger.js';
+import { LlmRole } from '../../telemetry/llmRole.js';
 
 export interface RollingSummaryProcessorOptions extends BackstopTargetOptions {
   systemInstruction?: string;
@@ -22,7 +22,42 @@ export function createRollingSummaryProcessor(
   env: ContextEnvironment,
   options: RollingSummaryProcessorOptions,
 ): ContextProcessor {
-  const generator = new SnapshotGenerator(env);
+  const generateRollingSummary = async (
+    nodes: ConcreteNode[],
+  ): Promise<string> => {
+    let transcript = '';
+    for (const node of nodes) {
+      let nodeContent = '';
+      if ('text' in node && typeof node.text === 'string') {
+        nodeContent = node.text;
+      } else if ('semanticParts' in node) {
+        nodeContent = JSON.stringify(node.semanticParts);
+      } else if ('observation' in node) {
+        nodeContent =
+          typeof node.observation === 'string'
+            ? node.observation
+            : JSON.stringify(node.observation);
+      }
+      transcript += `[${node.type}]: ${nodeContent}\n`;
+    }
+
+    const systemPrompt =
+      options.systemInstruction ??
+      `You are an expert context compressor. Your job is to drastically shorten the provided conversational transcript while preserving the absolute core semantic meaning, facts, and intent. Omit all conversational filler, pleasantries, or redundant information. Return ONLY the compressed summary.`;
+
+    const response = await env.llmClient.generateContent({
+      role: LlmRole.UTILITY_COMPRESSOR,
+      modelConfigKey: { model: 'gemini-3-flash-base' },
+      promptId: env.promptId,
+      abortSignal: new AbortController().signal,
+      contents: [{ role: 'user', parts: [{ text: transcript }] }],
+      systemInstruction: { role: 'system', parts: [{ text: systemPrompt }] },
+    });
+
+    const candidate = response.candidates?.[0];
+    const textPart = candidate?.content?.parts?.[0];
+    return textPart?.text || '';
+  };
 
   return {
     id,
@@ -66,10 +101,7 @@ export function createRollingSummaryProcessor(
 
       try {
         // Synthesize the rolling summary synchronously
-        const snapshotText = await generator.synthesizeSnapshot(
-          nodesToSummarize,
-          options.systemInstruction,
-        );
+        const snapshotText = await generateRollingSummary(nodesToSummarize);
         const newId = env.idGenerator.generateId();
 
         const summaryNode: RollingSummary = {