chore(release): v0.21.0-preview.4

Co-authored-by: Abhi <43648792+abhipatel12@users.noreply.github.com>

package-lock.json

@@ -1,12 +1,12 @@
 {
   "name": "@google/gemini-cli",
-  "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+  "version": "0.21.0-preview.4",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
       "name": "@google/gemini-cli",
-      "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+      "version": "0.21.0-preview.4",
       "workspaces": [
         "packages/*"
       ],
@@ -17692,7 +17692,7 @@
     },
     "packages/a2a-server": {
       "name": "@google/gemini-cli-a2a-server",
-      "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+      "version": "0.21.0-preview.4",
       "dependencies": {
         "@a2a-js/sdk": "^0.3.2",
         "@google-cloud/storage": "^7.16.0",
@@ -18003,7 +18003,7 @@
     },
     "packages/cli": {
       "name": "@google/gemini-cli",
-      "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+      "version": "0.21.0-preview.4",
       "dependencies": {
         "@google/gemini-cli-core": "file:../core",
         "@google/genai": "1.30.0",
@@ -18104,7 +18104,7 @@
     },
     "packages/core": {
       "name": "@google/gemini-cli-core",
-      "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+      "version": "0.21.0-preview.4",
       "dependencies": {
         "@google-cloud/logging": "^11.2.1",
         "@google-cloud/opentelemetry-cloud-monitoring-exporter": "^0.21.0",
@@ -18248,7 +18248,7 @@
     },
     "packages/test-utils": {
       "name": "@google/gemini-cli-test-utils",
-      "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+      "version": "0.21.0-preview.4",
       "license": "Apache-2.0",
       "devDependencies": {
         "typescript": "^5.3.3"
@@ -18259,7 +18259,7 @@
     },
     "packages/vscode-ide-companion": {
       "name": "gemini-cli-vscode-ide-companion",
-      "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+      "version": "0.21.0-preview.4",
       "license": "LICENSE",
       "dependencies": {
         "@modelcontextprotocol/sdk": "^1.23.0",

package.json

@@ -1,6 +1,6 @@
 {
   "name": "@google/gemini-cli",
-  "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+  "version": "0.21.0-preview.4",
   "engines": {
     "node": ">=20.0.0"
   },
@@ -14,7 +14,7 @@
     "url": "git+https://github.com/google-gemini/gemini-cli.git"
   },
   "config": {
-    "sandboxImageUri": "us-docker.pkg.dev/gemini-code-dev/gemini-cli/sandbox:0.21.0-nightly.20251209.ec9a8c7a7"
+    "sandboxImageUri": "us-docker.pkg.dev/gemini-code-dev/gemini-cli/sandbox:0.21.0-preview.4"
   },
   "scripts": {
     "start": "cross-env NODE_ENV=development node scripts/start.js",

packages/a2a-server/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@google/gemini-cli-a2a-server",
-  "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+  "version": "0.21.0-preview.4",
   "description": "Gemini CLI A2A Server",
   "repository": {
     "type": "git",

packages/cli/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@google/gemini-cli",
-  "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+  "version": "0.21.0-preview.4",
   "description": "Gemini CLI",
   "repository": {
     "type": "git",
@@ -25,7 +25,7 @@
     "dist"
   ],
   "config": {
-    "sandboxImageUri": "us-docker.pkg.dev/gemini-code-dev/gemini-cli/sandbox:0.21.0-nightly.20251209.ec9a8c7a7"
+    "sandboxImageUri": "us-docker.pkg.dev/gemini-code-dev/gemini-cli/sandbox:0.21.0-preview.4"
   },
   "dependencies": {
     "@google/gemini-cli-core": "file:../core",

packages/cli/src/gemini.tsx

@@ -498,6 +498,20 @@ export async function main() {
 
     // Handle --list-sessions flag
     if (config.getListSessions()) {
+      // Attempt auth for summary generation (gracefully skips if not configured)
+      const authType = settings.merged.security?.auth?.selectedType;
+      if (authType) {
+        try {
+          await config.refreshAuth(authType);
+        } catch (e) {
+          // Auth failed - continue without summary generation capability
+          debugLogger.debug(
+            'Auth failed for --list-sessions, summaries may not be generated:',
+            e,
+          );
+        }
+      }
+
       await listSessions(config);
       await runExitCleanup();
       process.exit(ExitCodes.SUCCESS);

packages/cli/src/ui/AppContainer.tsx

@@ -63,7 +63,7 @@ import {
   SessionEndReason,
   fireSessionStartHook,
   fireSessionEndHook,
-  generateAndSaveSummary,
+  generateSummary,
 } from '@google/gemini-cli-core';
 import { validateAuthMethod } from '../config/auth.js';
 import process from 'node:process';
@@ -312,9 +312,13 @@ export const AppContainer = (props: AppContainerProps) => {
           : SessionStartSource.Startup;
         await fireSessionStartHook(hookMessageBus, sessionStartSource);
       }
+
+      // Fire-and-forget: generate summary for previous session in background
+      generateSummary(config).catch((e) => {
+        debugLogger.warn('Background summary generation failed:', e);
+      });
     })();
     registerCleanup(async () => {
-      await generateAndSaveSummary(config);
       // Turn off mouse scroll.
       disableMouseEvents();
       const ideClient = await IdeClient.getInstance();

packages/cli/src/ui/components/ModelStatsDisplay.test.tsx

@@ -116,7 +116,7 @@ describe('<ModelStatsDisplay />', () => {
     });
 
     const output = lastFrame();
-    expect(output).not.toContain('Cached');
+    expect(output).not.toContain('Cache Reads');
     expect(output).not.toContain('Thoughts');
     expect(output).not.toContain('Tool');
     expect(output).toMatchSnapshot();
@@ -168,7 +168,7 @@ describe('<ModelStatsDisplay />', () => {
     });
 
     const output = lastFrame();
-    expect(output).toContain('Cached');
+    expect(output).toContain('Cache Reads');
     expect(output).toContain('Thoughts');
     expect(output).toContain('Tool');
     expect(output).toMatchSnapshot();

packages/cli/src/ui/components/ModelStatsDisplay.tsx

@@ -160,24 +160,28 @@ export const ModelStatsDisplay: React.FC = () => {
       <StatRow
         title="Total"
         values={getModelValues((m) => (
-          <Text color={theme.status.warning}>
+          <Text color={theme.text.secondary}>
             {m.tokens.total.toLocaleString()}
           </Text>
         ))}
       />
       <StatRow
-        title="Prompt"
+        title="Input"
         isSubtle
-        values={getModelValues((m) => m.tokens.prompt.toLocaleString())}
+        values={getModelValues((m) => (
+          <Text color={theme.text.primary}>
+            {Math.max(0, m.tokens.prompt - m.tokens.cached).toLocaleString()}
+          </Text>
+        ))}
       />
       {hasCached && (
         <StatRow
-          title="Cached"
+          title="Cache Reads"
           isSubtle
           values={getModelValues((m) => {
             const cacheHitRate = calculateCacheHitRate(m);
             return (
-              <Text color={theme.status.success}>
+              <Text color={theme.text.secondary}>
                 {m.tokens.cached.toLocaleString()} ({cacheHitRate.toFixed(1)}%)
               </Text>
             );
@@ -188,20 +192,32 @@ export const ModelStatsDisplay: React.FC = () => {
         <StatRow
           title="Thoughts"
           isSubtle
-          values={getModelValues((m) => m.tokens.thoughts.toLocaleString())}
+          values={getModelValues((m) => (
+            <Text color={theme.text.primary}>
+              {m.tokens.thoughts.toLocaleString()}
+            </Text>
+          ))}
         />
       )}
       {hasTool && (
         <StatRow
           title="Tool"
           isSubtle
-          values={getModelValues((m) => m.tokens.tool.toLocaleString())}
+          values={getModelValues((m) => (
+            <Text color={theme.text.primary}>
+              {m.tokens.tool.toLocaleString()}
+            </Text>
+          ))}
         />
       )}
       <StatRow
         title="Output"
         isSubtle
-        values={getModelValues((m) => m.tokens.candidates.toLocaleString())}
+        values={getModelValues((m) => (
+          <Text color={theme.text.primary}>
+            {m.tokens.candidates.toLocaleString()}
+          </Text>
+        ))}
       />
     </Box>
   );

packages/cli/src/ui/components/StatsDisplay.test.tsx

@@ -76,8 +76,6 @@ describe('<StatsDisplay />', () => {
 
     expect(output).toContain('Performance');
     expect(output).toContain('Interaction Summary');
-    expect(output).not.toContain('Efficiency & Optimizations');
-    expect(output).not.toContain('Model'); // The table header
     expect(output).toMatchSnapshot();
   });
 
@@ -114,8 +112,8 @@ describe('<StatsDisplay />', () => {
 
     expect(output).toContain('gemini-2.5-pro');
     expect(output).toContain('gemini-2.5-flash');
-    expect(output).toContain('1,000');
-    expect(output).toContain('25,000');
+    expect(output).toContain('15,000');
+    expect(output).toContain('10,000');
     expect(output).toMatchSnapshot();
   });
 
@@ -168,7 +166,6 @@ describe('<StatsDisplay />', () => {
     expect(output).toContain('Performance');
     expect(output).toContain('Interaction Summary');
     expect(output).toContain('User Agreement');
-    expect(output).toContain('Savings Highlight');
     expect(output).toContain('gemini-2.5-pro');
     expect(output).toMatchSnapshot();
   });
@@ -233,7 +230,6 @@ describe('<StatsDisplay />', () => {
       const { lastFrame } = renderWithMockedStats(metrics);
       const output = lastFrame();
 
-      expect(output).not.toContain('Efficiency & Optimizations');
       expect(output).toMatchSnapshot();
     });
   });
@@ -443,7 +439,7 @@ describe('<StatsDisplay />', () => {
       );
       const output = lastFrame();
 
-      expect(output).toContain('Usage limit remaining');
+      expect(output).toContain('Usage left');
       expect(output).toContain('75.0%');
       expect(output).toContain('(Resets in 1h 30m)');
       expect(output).toMatchSnapshot();

packages/cli/src/ui/components/StatsDisplay.tsx

@@ -17,9 +17,24 @@ import {
   TOOL_SUCCESS_RATE_MEDIUM,
   USER_AGREEMENT_RATE_HIGH,
   USER_AGREEMENT_RATE_MEDIUM,
+  CACHE_EFFICIENCY_HIGH,
+  CACHE_EFFICIENCY_MEDIUM,
 } from '../utils/displayUtils.js';
 import { computeSessionStats } from '../utils/computeStats.js';
 import type { RetrieveUserQuotaResponse } from '@google/gemini-cli-core';
+import {
+  DEFAULT_GEMINI_FLASH_LITE_MODEL,
+  DEFAULT_GEMINI_FLASH_MODEL,
+  DEFAULT_GEMINI_MODEL,
+  PREVIEW_GEMINI_MODEL,
+} from '@google/gemini-cli-core';
+
+const VALID_GEMINI_MODELS = new Set([
+  PREVIEW_GEMINI_MODEL,
+  DEFAULT_GEMINI_MODEL,
+  DEFAULT_GEMINI_FLASH_MODEL,
+  DEFAULT_GEMINI_FLASH_LITE_MODEL,
+]);
 
 // A more flexible and powerful StatRow component
 interface StatRowProps {
@@ -33,8 +48,7 @@ const StatRow: React.FC<StatRowProps> = ({ title, children }) => (
     <Box width={28}>
       <Text color={theme.text.link}>{title}</Text>
     </Box>
-    {/* FIX: Wrap children in a Box that can grow to fill remaining space */}
-    <Box flexGrow={1}>{children}</Box>
+    {children}
   </Box>
 );
 
@@ -50,8 +64,7 @@ const SubStatRow: React.FC<SubStatRowProps> = ({ title, children }) => (
     <Box width={26}>
       <Text color={theme.text.secondary}>» {title}</Text>
     </Box>
-    {/* FIX: Apply the same flexGrow fix here */}
-    <Box flexGrow={1}>{children}</Box>
+    {children}
   </Box>
 );
 
@@ -70,6 +83,57 @@ const Section: React.FC<SectionProps> = ({ title, children }) => (
   </Box>
 );
 
+// Logic for building the unified list of table rows
+const buildModelRows = (
+  models: Record<string, ModelMetrics>,
+  quotas?: RetrieveUserQuotaResponse,
+) => {
+  const getBaseModelName = (name: string) => name.replace('-001', '');
+  const usedModelNames = new Set(Object.keys(models).map(getBaseModelName));
+
+  // 1. Models with active usage
+  const activeRows = Object.entries(models).map(([name, metrics]) => {
+    const modelName = getBaseModelName(name);
+    const cachedTokens = metrics.tokens.cached;
+    const totalInputTokens = metrics.tokens.prompt;
+    const uncachedTokens = Math.max(0, totalInputTokens - cachedTokens);
+    return {
+      key: name,
+      modelName,
+      requests: metrics.api.totalRequests,
+      cachedTokens: cachedTokens.toLocaleString(),
+      uncachedTokens: uncachedTokens.toLocaleString(),
+      totalInputTokens: totalInputTokens.toLocaleString(),
+      outputTokens: metrics.tokens.candidates.toLocaleString(),
+      bucket: quotas?.buckets?.find((b) => b.modelId === modelName),
+      isActive: true,
+    };
+  });
+
+  // 2. Models with quota only
+  const quotaRows =
+    quotas?.buckets
+      ?.filter(
+        (b) =>
+          b.modelId &&
+          VALID_GEMINI_MODELS.has(b.modelId) &&
+          !usedModelNames.has(b.modelId),
+      )
+      .map((bucket) => ({
+        key: bucket.modelId!,
+        modelName: bucket.modelId!,
+        requests: '-',
+        cachedTokens: '-',
+        uncachedTokens: '-',
+        totalInputTokens: '-',
+        outputTokens: '-',
+        bucket,
+        isActive: false,
+      })) || [];
+
+  return [...activeRows, ...quotaRows];
+};
+
 const formatResetTime = (resetTime: string): string => {
   const diff = new Date(resetTime).getTime() - Date.now();
   if (diff <= 0) return '';
@@ -96,44 +160,91 @@ const formatResetTime = (resetTime: string): string => {
 
 const ModelUsageTable: React.FC<{
   models: Record<string, ModelMetrics>;
-  totalCachedTokens: number;
-  cacheEfficiency: number;
   quotas?: RetrieveUserQuotaResponse;
-}> = ({ models, totalCachedTokens, cacheEfficiency, quotas }) => {
+  cacheEfficiency: number;
+  totalCachedTokens: number;
+}> = ({ models, quotas, cacheEfficiency, totalCachedTokens }) => {
+  const rows = buildModelRows(models, quotas);
+
+  if (rows.length === 0) {
+    return null;
+  }
+
+  const showQuotaColumn = !!quotas && rows.some((row) => !!row.bucket);
+
   const nameWidth = 25;
-  const requestsWidth = 8;
-  const inputTokensWidth = 15;
+  const requestsWidth = 7;
+  const uncachedWidth = 15;
+  const cachedWidth = 14;
   const outputTokensWidth = 15;
-  const usageLimitWidth = quotas ? 30 : 0;
+  const usageLimitWidth = showQuotaColumn ? 28 : 0;
+
+  const cacheEfficiencyColor = getStatusColor(cacheEfficiency, {
+    green: CACHE_EFFICIENCY_HIGH,
+    yellow: CACHE_EFFICIENCY_MEDIUM,
+  });
 
   return (
     <Box flexDirection="column" marginTop={1}>
       {/* Header */}
-      <Box>
-        <Box width={nameWidth}>
-          <Text bold color={theme.text.primary}>
+      <Box alignItems="flex-end">
+        <Box width={nameWidth} flexGrow={1}>
+          <Text bold color={theme.text.primary} wrap="truncate-end">
             Model Usage
           </Text>
         </Box>
-        <Box width={requestsWidth} justifyContent="flex-end">
+        <Box
+          width={requestsWidth}
+          flexDirection="column"
+          alignItems="flex-end"
+          flexShrink={0}
+        >
           <Text bold color={theme.text.primary}>
             Reqs
           </Text>
         </Box>
-        <Box width={inputTokensWidth} justifyContent="flex-end">
-          <Text bold color={theme.text.primary}>
-            Input Tokens
-          </Text>
-        </Box>
-        <Box width={outputTokensWidth} justifyContent="flex-end">
-          <Text bold color={theme.text.primary}>
-            Output Tokens
-          </Text>
-        </Box>
-        {quotas && (
-          <Box width={usageLimitWidth} justifyContent="flex-end">
+        {!showQuotaColumn && (
+          <>
+            <Box
+              width={uncachedWidth}
+              flexDirection="column"
+              alignItems="flex-end"
+              flexShrink={0}
+            >
+              <Text bold color={theme.text.primary}>
+                Input Tokens
+              </Text>
+            </Box>
+            <Box
+              width={cachedWidth}
+              flexDirection="column"
+              alignItems="flex-end"
+              flexShrink={0}
+            >
+              <Text bold color={theme.text.primary}>
+                Cache Reads
+              </Text>
+            </Box>
+            <Box
+              width={outputTokensWidth}
+              flexDirection="column"
+              alignItems="flex-end"
+              flexShrink={0}
+            >
+              <Text bold color={theme.text.primary}>
+                Output Tokens
+              </Text>
+            </Box>
+          </>
+        )}
+        {showQuotaColumn && (
+          <Box
+            width={usageLimitWidth}
+            flexDirection="column"
+            alignItems="flex-end"
+          >
             <Text bold color={theme.text.primary}>
-              Usage limit remaining
+              Usage left
             </Text>
           </Box>
         )}
@@ -146,63 +257,99 @@ const ModelUsageTable: React.FC<{
         borderLeft={false}
         borderRight={false}
         borderColor={theme.border.default}
-        width={
-          nameWidth +
-          requestsWidth +
-          inputTokensWidth +
-          outputTokensWidth +
-          usageLimitWidth
-        }
+        width="100%"
       ></Box>
 
-      {/* Rows */}
-      {Object.entries(models).map(([name, modelMetrics]) => {
-        const modelName = name.replace('-001', '');
-        const bucket = quotas?.buckets?.find((b) => b.modelId === modelName);
-
-        return (
-          <Box key={name}>
-            <Box width={nameWidth}>
-              <Text color={theme.text.primary}>{modelName}</Text>
-            </Box>
-            <Box width={requestsWidth} justifyContent="flex-end">
-              <Text color={theme.text.primary}>
-                {modelMetrics.api.totalRequests}
-              </Text>
-            </Box>
-            <Box width={inputTokensWidth} justifyContent="flex-end">
-              <Text color={theme.status.warning}>
-                {modelMetrics.tokens.prompt.toLocaleString()}
-              </Text>
-            </Box>
-            <Box width={outputTokensWidth} justifyContent="flex-end">
-              <Text color={theme.status.warning}>
-                {modelMetrics.tokens.candidates.toLocaleString()}
-              </Text>
-            </Box>
-            <Box width={usageLimitWidth} justifyContent="flex-end">
-              {bucket &&
-                bucket.remainingFraction != null &&
-                bucket.resetTime && (
-                  <Text color={theme.text.secondary}>
-                    {(bucket.remainingFraction * 100).toFixed(1)}%{' '}
-                    {formatResetTime(bucket.resetTime)}
-                  </Text>
-                )}
-            </Box>
+      {rows.map((row) => (
+        <Box key={row.key}>
+          <Box width={nameWidth} flexGrow={1}>
+            <Text color={theme.text.primary} wrap="truncate-end">
+              {row.modelName}
+            </Text>
           </Box>
-        );
-      })}
-      {cacheEfficiency > 0 && (
+          <Box
+            width={requestsWidth}
+            flexDirection="column"
+            alignItems="flex-end"
+            flexShrink={0}
+          >
+            <Text
+              color={row.isActive ? theme.text.primary : theme.text.secondary}
+            >
+              {row.requests}
+            </Text>
+          </Box>
+          {!showQuotaColumn && (
+            <>
+              <Box
+                width={uncachedWidth}
+                flexDirection="column"
+                alignItems="flex-end"
+                flexShrink={0}
+              >
+                <Text
+                  color={
+                    row.isActive ? theme.text.primary : theme.text.secondary
+                  }
+                >
+                  {row.uncachedTokens}
+                </Text>
+              </Box>
+              <Box
+                width={cachedWidth}
+                flexDirection="column"
+                alignItems="flex-end"
+                flexShrink={0}
+              >
+                <Text color={theme.text.secondary}>{row.cachedTokens}</Text>
+              </Box>
+              <Box
+                width={outputTokensWidth}
+                flexDirection="column"
+                alignItems="flex-end"
+                flexShrink={0}
+              >
+                <Text
+                  color={
+                    row.isActive ? theme.text.primary : theme.text.secondary
+                  }
+                >
+                  {row.outputTokens}
+                </Text>
+              </Box>
+            </>
+          )}
+          <Box
+            width={usageLimitWidth}
+            flexDirection="column"
+            alignItems="flex-end"
+          >
+            {row.bucket &&
+              row.bucket.remainingFraction != null &&
+              row.bucket.resetTime && (
+                <Text color={theme.text.secondary} wrap="truncate-end">
+                  {(row.bucket.remainingFraction * 100).toFixed(1)}%{' '}
+                  {formatResetTime(row.bucket.resetTime)}
+                </Text>
+              )}
+          </Box>
+        </Box>
+      ))}
+
+      {cacheEfficiency > 0 && !showQuotaColumn && (
         <Box flexDirection="column" marginTop={1}>
           <Text color={theme.text.primary}>
             <Text color={theme.status.success}>Savings Highlight:</Text>{' '}
-            {totalCachedTokens.toLocaleString()} ({cacheEfficiency.toFixed(1)}
-            %) of input tokens were served from the cache, reducing costs.
+            {totalCachedTokens.toLocaleString()} (
+            <Text color={cacheEfficiencyColor}>
+              {cacheEfficiency.toFixed(1)}%
+            </Text>
+            ) of input tokens were served from the cache, reducing costs.
           </Text>
         </Box>
       )}
-      {models && (
+
+      {showQuotaColumn && (
         <>
           <Box marginTop={1} marginBottom={2}>
             <Text color={theme.text.primary}>
@@ -266,6 +413,7 @@ export const StatsDisplay: React.FC<StatsDisplayProps> = ({
       flexDirection="column"
       paddingY={1}
       paddingX={2}
+      overflow="hidden"
     >
       {renderTitle()}
       <Box height={1} />
@@ -335,15 +483,12 @@ export const StatsDisplay: React.FC<StatsDisplayProps> = ({
           </Text>
         </SubStatRow>
       </Section>
-
-      {Object.keys(models).length > 0 && (
-        <ModelUsageTable
-          models={models}
-          totalCachedTokens={computed.totalCachedTokens}
-          cacheEfficiency={computed.cacheEfficiency}
-          quotas={quotas}
-        />
-      )}
+      <ModelUsageTable
+        models={models}
+        quotas={quotas}
+        cacheEfficiency={computed.cacheEfficiency}
+        totalCachedTokens={computed.totalCachedTokens}
+      />
     </Box>
   );
 };

packages/cli/src/ui/components/__snapshots__/ModelStatsDisplay.test.tsx.snap

@@ -14,8 +14,8 @@ exports[`<ModelStatsDisplay /> > should display a single model correctly 1`] = `
 │                                                                                                  │
 │  Tokens                                                                                          │
 │  Total                       30                                                                  │
-│    ↳ Prompt                  10                                                                  │
-│    ↳ Cached                  5 (50.0%)                                                           │
+│    ↳ Input                   5                                                                   │
+│    ↳ Cache Reads             5 (50.0%)                                                           │
 │    ↳ Thoughts                2                                                                   │
 │    ↳ Tool                    1                                                                   │
 │    ↳ Output                  20                                                                  │
@@ -37,8 +37,8 @@ exports[`<ModelStatsDisplay /> > should display conditional rows if at least one
 │                                                                                                  │
 │  Tokens                                                                                          │
 │  Total                       30                    15                                            │
-│    ↳ Prompt                  10                    5                                             │
-│    ↳ Cached                  5 (50.0%)             0 (0.0%)                                      │
+│    ↳ Input                   5                     5                                             │
+│    ↳ Cache Reads             5 (50.0%)             0 (0.0%)                                      │
 │    ↳ Thoughts                2                     0                                             │
 │    ↳ Tool                    0                     3                                             │
 │    ↳ Output                  20                    10                                            │
@@ -60,8 +60,8 @@ exports[`<ModelStatsDisplay /> > should display stats for multiple models correc
 │                                                                                                  │
 │  Tokens                                                                                          │
 │  Total                       300                   600                                           │
-│    ↳ Prompt                  100                   200                                           │
-│    ↳ Cached                  50 (50.0%)            100 (50.0%)                                   │
+│    ↳ Input                   50                    100                                           │
+│    ↳ Cache Reads             50 (50.0%)            100 (50.0%)                                   │
 │    ↳ Thoughts                10                    20                                            │
 │    ↳ Tool                    5                     10                                            │
 │    ↳ Output                  200                   400                                           │
@@ -83,8 +83,8 @@ exports[`<ModelStatsDisplay /> > should handle large values without wrapping or
 │                                                                                                  │
 │  Tokens                                                                                          │
 │  Total                       999,999,999                                                         │
-│    ↳ Prompt                  987,654,321                                                         │
-│    ↳ Cached                  123,456,789 (12.5%)                                                 │
+│    ↳ Input                   864,197,532                                                         │
+│    ↳ Cache Reads             123,456,789 (12.5%)                                                 │
 │    ↳ Thoughts                111,111,111                                                         │
 │    ↳ Tool                    222,222,222                                                         │
 │    ↳ Output                  123,456,789                                                         │
@@ -106,7 +106,7 @@ exports[`<ModelStatsDisplay /> > should not display conditional rows if no model
 │                                                                                                  │
 │  Tokens                                                                                          │
 │  Total                       30                                                                  │
-│    ↳ Prompt                  10                                                                  │
+│    ↳ Input                   10                                                                  │
 │    ↳ Output                  20                                                                  │
 │                                                                                                  │
 ╰──────────────────────────────────────────────────────────────────────────────────────────────────╯"

packages/cli/src/ui/components/__snapshots__/SessionSummaryDisplay.test.tsx.snap

@@ -18,17 +18,11 @@ exports[`<SessionSummaryDisplay /> > renders the summary display with a title 1`
 │    » Tool Time:              0s (0.0%)                                                           │
 │                                                                                                  │
 │                                                                                                  │
-│  Model Usage                  Reqs   Input Tokens  Output Tokens                                 │
-│  ───────────────────────────────────────────────────────────────                                 │
-│  gemini-2.5-pro                 10          1,000          2,000                                 │
+│  Model Usage                                   Reqs   Input Tokens   Cache Reads  Output Tokens  │
+│  ──────────────────────────────────────────────────────────────────────────────────────────────  │
+│  gemini-2.5-pro                                  10            500           500          2,000  │
 │                                                                                                  │
 │  Savings Highlight: 500 (50.0%) of input tokens were served from the cache, reducing costs.      │
 │                                                                                                  │
-│  Usage limits span all sessions and reset daily.                                                 │
-│  /auth to upgrade or switch to API key.                                                          │
-│                                                                                                  │
-│                                                                                                  │
-│  » Tip: For a full token breakdown, run \`/stats model\`.                                          │
-│                                                                                                  │
 ╰──────────────────────────────────────────────────────────────────────────────────────────────────╯"
 `;

packages/cli/src/ui/components/__snapshots__/StatsDisplay.test.tsx.snap

@@ -118,15 +118,9 @@ exports[`<StatsDisplay /> > Conditional Rendering Tests > hides Efficiency secti
 │    » Tool Time:              0s (0.0%)                                                           │
 │                                                                                                  │
 │                                                                                                  │
-│  Model Usage                  Reqs   Input Tokens  Output Tokens                                 │
-│  ───────────────────────────────────────────────────────────────                                 │
-│  gemini-2.5-pro                  1            100            100                                 │
-│                                                                                                  │
-│  Usage limits span all sessions and reset daily.                                                 │
-│  /auth to upgrade or switch to API key.                                                          │
-│                                                                                                  │
-│                                                                                                  │
-│  » Tip: For a full token breakdown, run \`/stats model\`.                                          │
+│  Model Usage                                   Reqs   Input Tokens   Cache Reads  Output Tokens  │
+│  ──────────────────────────────────────────────────────────────────────────────────────────────  │
+│  gemini-2.5-pro                                   1            100             0            100  │
 │                                                                                                  │
 ╰──────────────────────────────────────────────────────────────────────────────────────────────────╯"
 `;
@@ -168,11 +162,9 @@ exports[`<StatsDisplay /> > Quota Display > renders quota information when quota
 │    » Tool Time:              0s (0.0%)                                                           │
 │                                                                                                  │
 │                                                                                                  │
-│  Model Usage                  Reqs   Input Tokens  Output Tokens         Usage limit remaining   │
-│  ─────────────────────────────────────────────────────────────────────────────────────────────   │
-│  gemini-2.5-pro                  1            100            100      75.0% (Resets in 1h 30m)   │
-│                                                                                                  │
-│  Savings Highlight: 50 (50.0%) of input tokens were served from the cache, reducing costs.       │
+│  Model Usage                                                   Reqs                  Usage left  │
+│  ──────────────────────────────────────────────────────────────────────────────────────────────  │
+│  gemini-2.5-pro                                                   1    75.0% (Resets in 1h 30m)  │
 │                                                                                                  │
 │  Usage limits span all sessions and reset daily.                                                 │
 │  /auth to upgrade or switch to API key.                                                          │
@@ -240,19 +232,13 @@ exports[`<StatsDisplay /> > renders a table with two models correctly 1`] = `
 │    » Tool Time:              0s (0.0%)                                                           │
 │                                                                                                  │
 │                                                                                                  │
-│  Model Usage                  Reqs   Input Tokens  Output Tokens                                 │
-│  ───────────────────────────────────────────────────────────────                                 │
-│  gemini-2.5-pro                  3          1,000          2,000                                 │
-│  gemini-2.5-flash                5         25,000         15,000                                 │
+│  Model Usage                                   Reqs   Input Tokens   Cache Reads  Output Tokens  │
+│  ──────────────────────────────────────────────────────────────────────────────────────────────  │
+│  gemini-2.5-pro                                   3            500           500          2,000  │
+│  gemini-2.5-flash                                 5         15,000        10,000         15,000  │
 │                                                                                                  │
 │  Savings Highlight: 10,500 (40.4%) of input tokens were served from the cache, reducing costs.   │
 │                                                                                                  │
-│  Usage limits span all sessions and reset daily.                                                 │
-│  /auth to upgrade or switch to API key.                                                          │
-│                                                                                                  │
-│                                                                                                  │
-│  » Tip: For a full token breakdown, run \`/stats model\`.                                          │
-│                                                                                                  │
 ╰──────────────────────────────────────────────────────────────────────────────────────────────────╯"
 `;
 
@@ -274,18 +260,12 @@ exports[`<StatsDisplay /> > renders all sections when all data is present 1`] =
 │    » Tool Time:              123ms (55.2%)                                                       │
 │                                                                                                  │
 │                                                                                                  │
-│  Model Usage                  Reqs   Input Tokens  Output Tokens                                 │
-│  ───────────────────────────────────────────────────────────────                                 │
-│  gemini-2.5-pro                  1            100            100                                 │
+│  Model Usage                                   Reqs   Input Tokens   Cache Reads  Output Tokens  │
+│  ──────────────────────────────────────────────────────────────────────────────────────────────  │
+│  gemini-2.5-pro                                   1             50            50            100  │
 │                                                                                                  │
 │  Savings Highlight: 50 (50.0%) of input tokens were served from the cache, reducing costs.       │
 │                                                                                                  │
-│  Usage limits span all sessions and reset daily.                                                 │
-│  /auth to upgrade or switch to API key.                                                          │
-│                                                                                                  │
-│                                                                                                  │
-│  » Tip: For a full token breakdown, run \`/stats model\`.                                          │
-│                                                                                                  │
 ╰──────────────────────────────────────────────────────────────────────────────────────────────────╯"
 `;
 

packages/cli/src/utils/sessions.test.ts

@@ -21,6 +21,7 @@ vi.mock('@google/gemini-cli-core', async () => {
   return {
     ...actual,
     ChatRecordingService: vi.fn(),
+    generateSummary: vi.fn().mockResolvedValue(undefined),
   };
 });
 

packages/cli/src/utils/sessions.ts

@@ -4,7 +4,11 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import { ChatRecordingService, type Config } from '@google/gemini-cli-core';
+import {
+  ChatRecordingService,
+  generateSummary,
+  type Config,
+} from '@google/gemini-cli-core';
 import {
   formatRelativeTime,
   SessionSelector,
@@ -12,6 +16,9 @@ import {
 } from './sessionUtils.js';
 
 export async function listSessions(config: Config): Promise<void> {
+  // Generate summary for most recent session if needed
+  await generateSummary(config);
+
   const sessionSelector = new SessionSelector(config);
   const sessions = await sessionSelector.listSessions();
 

packages/core/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@google/gemini-cli-core",
-  "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+  "version": "0.21.0-preview.4",
   "description": "Gemini CLI Core",
   "repository": {
     "type": "git",

packages/core/src/ide/ide-client.test.ts

@@ -59,6 +59,7 @@ describe('IdeClient', () => {
     delete process.env['GEMINI_CLI_IDE_SERVER_PORT'];
     delete process.env['GEMINI_CLI_IDE_SERVER_STDIO_COMMAND'];
     delete process.env['GEMINI_CLI_IDE_SERVER_STDIO_ARGS'];
+    delete process.env['GEMINI_CLI_IDE_AUTH_TOKEN'];
 
     // Mock dependencies
     vi.spyOn(process, 'cwd').mockReturnValue('/test/workspace/sub-dir');
@@ -923,5 +924,35 @@ describe('IdeClient', () => {
         IDEConnectionStatus.Connected,
       );
     });
+
+    it('should connect with an auth token from environment variable if config file is missing', async () => {
+      vi.mocked(fs.promises.readFile).mockRejectedValue(
+        new Error('File not found'),
+      );
+      (
+        vi.mocked(fs.promises.readdir) as Mock<
+          (path: fs.PathLike) => Promise<string[]>
+        >
+      ).mockResolvedValue([]);
+      process.env['GEMINI_CLI_IDE_SERVER_PORT'] = '9090';
+      process.env['GEMINI_CLI_IDE_AUTH_TOKEN'] = 'env-auth-token';
+
+      const ideClient = await IdeClient.getInstance();
+      await ideClient.connect();
+
+      expect(StreamableHTTPClientTransport).toHaveBeenCalledWith(
+        new URL('http://127.0.0.1:9090/mcp'),
+        expect.objectContaining({
+          requestInit: {
+            headers: {
+              Authorization: 'Bearer env-auth-token',
+            },
+          },
+        }),
+      );
+      expect(ideClient.getConnectionStatus().status).toBe(
+        IDEConnectionStatus.Connected,
+      );
+    });
   });
 });

packages/core/src/ide/ide-client.ts

@@ -151,9 +151,10 @@ export class IdeClient {
     this.setState(IDEConnectionStatus.Connecting);
 
     this.connectionConfig = await this.getConnectionConfigFromFile();
-    if (this.connectionConfig?.authToken) {
-      this.authToken = this.connectionConfig.authToken;
-    }
+    this.authToken =
+      this.connectionConfig?.authToken ??
+      process.env['GEMINI_CLI_IDE_AUTH_TOKEN'];
+
     const workspacePath =
       this.connectionConfig?.workspacePath ??
       process.env['GEMINI_CLI_IDE_WORKSPACE_PATH'];

packages/core/src/policy/policies/read-only.toml

@@ -54,3 +54,8 @@ priority = 50
 toolName = "google_web_search"
 decision = "allow"
 priority = 50
+
+[[rule]]
+toolName = "codebase_investigator"
+decision = "allow"
+priority = 50

packages/core/src/services/sessionSummaryUtils.test.ts

@@ -5,11 +5,15 @@
  */
 
 import { describe, it, expect, vi, beforeEach, afterEach } from 'vitest';
-import { generateAndSaveSummary } from './sessionSummaryUtils.js';
+import { generateSummary, getPreviousSession } from './sessionSummaryUtils.js';
 import type { Config } from '../config/config.js';
-import type { ChatRecordingService } from './chatRecordingService.js';
 import type { ContentGenerator } from '../core/contentGenerator.js';
-import type { GeminiClient } from '../core/client.js';
+import * as fs from 'node:fs/promises';
+import * as path from 'node:path';
+
+// Mock fs/promises
+vi.mock('node:fs/promises');
+const mockReaddir = fs.readdir as unknown as ReturnType<typeof vi.fn>;
 
 // Mock the SessionSummaryService module
 vi.mock('./sessionSummaryService.js', () => ({
@@ -23,10 +27,24 @@ vi.mock('../core/baseLlmClient.js', () => ({
   BaseLlmClient: vi.fn(),
 }));
 
+// Helper to create a session with N user messages
+function createSessionWithUserMessages(
+  count: number,
+  options: { summary?: string; sessionId?: string } = {},
+) {
+  return JSON.stringify({
+    sessionId: options.sessionId ?? 'session-id',
+    summary: options.summary,
+    messages: Array.from({ length: count }, (_, i) => ({
+      id: String(i + 1),
+      type: 'user',
+      content: [{ text: `Message ${i + 1}` }],
+    })),
+  });
+}
+
 describe('sessionSummaryUtils', () => {
   let mockConfig: Config;
-  let mockChatRecordingService: ChatRecordingService;
-  let mockGeminiClient: GeminiClient;
   let mockContentGenerator: ContentGenerator;
   let mockGenerateSummary: ReturnType<typeof vi.fn>;
 
@@ -36,23 +54,12 @@ describe('sessionSummaryUtils', () => {
     // Setup mock content generator
     mockContentGenerator = {} as ContentGenerator;
 
-    // Setup mock chat recording service
-    mockChatRecordingService = {
-      getConversation: vi.fn(),
-      saveSummary: vi.fn(),
-    } as unknown as ChatRecordingService;
-
-    // Setup mock gemini client
-    mockGeminiClient = {
-      getChatRecordingService: vi
-        .fn()
-        .mockReturnValue(mockChatRecordingService),
-    } as unknown as GeminiClient;
-
     // Setup mock config
     mockConfig = {
       getContentGenerator: vi.fn().mockReturnValue(mockContentGenerator),
-      getGeminiClient: vi.fn().mockReturnValue(mockGeminiClient),
+      storage: {
+        getProjectTempDir: vi.fn().mockReturnValue('/tmp/project'),
+      },
     } as unknown as Config;
 
     // Setup mock generateSummary function
@@ -73,320 +80,138 @@ describe('sessionSummaryUtils', () => {
     vi.restoreAllMocks();
   });
 
-  describe('Integration Tests', () => {
-    it('should generate and save summary successfully', async () => {
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        messages: [
-          {
-            id: '1',
-            timestamp: '2025-12-03T00:00:00Z',
-            type: 'user' as const,
-            content: [{ text: 'How do I add dark mode?' }],
-          },
-          {
-            id: '2',
-            timestamp: '2025-12-03T00:01:00Z',
-            type: 'gemini' as const,
-            content: [{ text: 'To add dark mode...' }],
-          },
-        ],
-      };
+  describe('getPreviousSession', () => {
+    it('should return null if chats directory does not exist', async () => {
+      vi.mocked(fs.access).mockRejectedValue(new Error('ENOENT'));
 
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
+      const result = await getPreviousSession(mockConfig);
 
-      await generateAndSaveSummary(mockConfig);
+      expect(result).toBeNull();
+    });
 
-      expect(mockChatRecordingService.getConversation).toHaveBeenCalledTimes(1);
-      expect(mockGenerateSummary).toHaveBeenCalledTimes(1);
-      expect(mockGenerateSummary).toHaveBeenCalledWith({
-        messages: mockConversation.messages,
-      });
-      expect(mockChatRecordingService.saveSummary).toHaveBeenCalledTimes(1);
-      expect(mockChatRecordingService.saveSummary).toHaveBeenCalledWith(
-        'Add dark mode to the app',
+    it('should return null if no session files exist', async () => {
+      vi.mocked(fs.access).mockResolvedValue(undefined);
+      mockReaddir.mockResolvedValue([]);
+
+      const result = await getPreviousSession(mockConfig);
+
+      expect(result).toBeNull();
+    });
+
+    it('should return null if most recent session already has summary', async () => {
+      vi.mocked(fs.access).mockResolvedValue(undefined);
+      mockReaddir.mockResolvedValue(['session-2024-01-01T10-00-abc12345.json']);
+      vi.mocked(fs.readFile).mockResolvedValue(
+        createSessionWithUserMessages(5, { summary: 'Existing summary' }),
+      );
+
+      const result = await getPreviousSession(mockConfig);
+
+      expect(result).toBeNull();
+    });
+
+    it('should return null if most recent session has 1 or fewer user messages', async () => {
+      vi.mocked(fs.access).mockResolvedValue(undefined);
+      mockReaddir.mockResolvedValue(['session-2024-01-01T10-00-abc12345.json']);
+      vi.mocked(fs.readFile).mockResolvedValue(
+        createSessionWithUserMessages(1),
+      );
+
+      const result = await getPreviousSession(mockConfig);
+
+      expect(result).toBeNull();
+    });
+
+    it('should return path if most recent session has more than 1 user message and no summary', async () => {
+      vi.mocked(fs.access).mockResolvedValue(undefined);
+      mockReaddir.mockResolvedValue(['session-2024-01-01T10-00-abc12345.json']);
+      vi.mocked(fs.readFile).mockResolvedValue(
+        createSessionWithUserMessages(2),
+      );
+
+      const result = await getPreviousSession(mockConfig);
+
+      expect(result).toBe(
+        path.join(
+          '/tmp/project',
+          'chats',
+          'session-2024-01-01T10-00-abc12345.json',
+        ),
       );
     });
 
-    it('should skip if no chat recording service is available', async () => {
-      (
-        mockGeminiClient.getChatRecordingService as ReturnType<typeof vi.fn>
-      ).mockReturnValue(undefined);
+    it('should select most recently created session by filename', async () => {
+      vi.mocked(fs.access).mockResolvedValue(undefined);
+      mockReaddir.mockResolvedValue([
+        'session-2024-01-01T10-00-older000.json',
+        'session-2024-01-02T10-00-newer000.json',
+      ]);
+      vi.mocked(fs.readFile).mockResolvedValue(
+        createSessionWithUserMessages(2),
+      );
 
-      await generateAndSaveSummary(mockConfig);
+      const result = await getPreviousSession(mockConfig);
 
-      expect(mockGeminiClient.getChatRecordingService).toHaveBeenCalledTimes(1);
-      expect(mockGenerateSummary).not.toHaveBeenCalled();
-      expect(mockChatRecordingService.saveSummary).not.toHaveBeenCalled();
+      expect(result).toBe(
+        path.join(
+          '/tmp/project',
+          'chats',
+          'session-2024-01-02T10-00-newer000.json',
+        ),
+      );
     });
 
-    it('should skip if no conversation exists', async () => {
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(null);
+    it('should return null if most recent session file is corrupted', async () => {
+      vi.mocked(fs.access).mockResolvedValue(undefined);
+      mockReaddir.mockResolvedValue(['session-2024-01-01T10-00-abc12345.json']);
+      vi.mocked(fs.readFile).mockResolvedValue('invalid json');
 
-      await generateAndSaveSummary(mockConfig);
+      const result = await getPreviousSession(mockConfig);
 
-      expect(mockChatRecordingService.getConversation).toHaveBeenCalledTimes(1);
-      expect(mockGenerateSummary).not.toHaveBeenCalled();
-      expect(mockChatRecordingService.saveSummary).not.toHaveBeenCalled();
-    });
-
-    it('should skip if summary already exists', async () => {
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        summary: 'Existing summary',
-        messages: [
-          {
-            id: '1',
-            timestamp: '2025-12-03T00:00:00Z',
-            type: 'user' as const,
-            content: [{ text: 'Hello' }],
-          },
-        ],
-      };
-
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
-
-      await generateAndSaveSummary(mockConfig);
-
-      expect(mockChatRecordingService.getConversation).toHaveBeenCalledTimes(1);
-      expect(mockGenerateSummary).not.toHaveBeenCalled();
-      expect(mockChatRecordingService.saveSummary).not.toHaveBeenCalled();
-    });
-
-    it('should skip if no messages present', async () => {
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        messages: [],
-      };
-
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
-
-      await generateAndSaveSummary(mockConfig);
-
-      expect(mockChatRecordingService.getConversation).toHaveBeenCalledTimes(1);
-      expect(mockGenerateSummary).not.toHaveBeenCalled();
-      expect(mockChatRecordingService.saveSummary).not.toHaveBeenCalled();
-    });
-
-    it('should handle generateSummary failure gracefully', async () => {
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        messages: [
-          {
-            id: '1',
-            timestamp: '2025-12-03T00:00:00Z',
-            type: 'user' as const,
-            content: [{ text: 'Hello' }],
-          },
-        ],
-      };
-
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
-      mockGenerateSummary.mockResolvedValue(null);
-
-      await generateAndSaveSummary(mockConfig);
-
-      expect(mockChatRecordingService.getConversation).toHaveBeenCalledTimes(1);
-      expect(mockGenerateSummary).toHaveBeenCalledTimes(1);
-      expect(mockChatRecordingService.saveSummary).not.toHaveBeenCalled();
-    });
-
-    it('should catch and log errors without throwing', async () => {
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        messages: [
-          {
-            id: '1',
-            timestamp: '2025-12-03T00:00:00Z',
-            type: 'user' as const,
-            content: [{ text: 'Hello' }],
-          },
-        ],
-      };
-
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
-      mockGenerateSummary.mockRejectedValue(new Error('API Error'));
-
-      // Should not throw
-      await expect(generateAndSaveSummary(mockConfig)).resolves.not.toThrow();
-
-      expect(mockChatRecordingService.getConversation).toHaveBeenCalledTimes(1);
-      expect(mockGenerateSummary).toHaveBeenCalledTimes(1);
-      expect(mockChatRecordingService.saveSummary).not.toHaveBeenCalled();
+      expect(result).toBeNull();
     });
   });
 
-  describe('Mock Verification Tests', () => {
-    it('should call getConversation() once', async () => {
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        messages: [
-          {
-            id: '1',
-            timestamp: '2025-12-03T00:00:00Z',
-            type: 'user' as const,
-            content: [{ text: 'Hello' }],
-          },
-        ],
-      };
+  describe('generateSummary', () => {
+    it('should not throw if getPreviousSession returns null', async () => {
+      vi.mocked(fs.access).mockRejectedValue(new Error('ENOENT'));
 
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
-
-      await generateAndSaveSummary(mockConfig);
-
-      expect(mockChatRecordingService.getConversation).toHaveBeenCalledTimes(1);
-      expect(mockChatRecordingService.getConversation).toHaveBeenCalledWith();
+      await expect(generateSummary(mockConfig)).resolves.not.toThrow();
     });
 
-    it('should call generateSummary() with correct messages', async () => {
-      const mockMessages = [
-        {
-          id: '1',
-          timestamp: '2025-12-03T00:00:00Z',
-          type: 'user' as const,
-          content: [{ text: 'How do I add dark mode?' }],
-        },
-        {
-          id: '2',
-          timestamp: '2025-12-03T00:01:00Z',
-          type: 'gemini' as const,
-          content: [{ text: 'To add dark mode...' }],
-        },
-      ];
+    it('should generate and save summary for session needing one', async () => {
+      const sessionPath = path.join(
+        '/tmp/project',
+        'chats',
+        'session-2024-01-01T10-00-abc12345.json',
+      );
 
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        messages: mockMessages,
-      };
+      vi.mocked(fs.access).mockResolvedValue(undefined);
+      mockReaddir.mockResolvedValue(['session-2024-01-01T10-00-abc12345.json']);
+      vi.mocked(fs.readFile).mockResolvedValue(
+        createSessionWithUserMessages(2),
+      );
+      vi.mocked(fs.writeFile).mockResolvedValue(undefined);
 
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
-
-      await generateAndSaveSummary(mockConfig);
+      await generateSummary(mockConfig);
 
       expect(mockGenerateSummary).toHaveBeenCalledTimes(1);
-      expect(mockGenerateSummary).toHaveBeenCalledWith({
-        messages: mockMessages,
-      });
-    });
-
-    it('should call saveSummary() with generated summary', async () => {
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        messages: [
-          {
-            id: '1',
-            timestamp: '2025-12-03T00:00:00Z',
-            type: 'user' as const,
-            content: [{ text: 'Hello' }],
-          },
-        ],
-      };
-
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
-      mockGenerateSummary.mockResolvedValue('Test summary');
-
-      await generateAndSaveSummary(mockConfig);
-
-      expect(mockChatRecordingService.saveSummary).toHaveBeenCalledTimes(1);
-      expect(mockChatRecordingService.saveSummary).toHaveBeenCalledWith(
-        'Test summary',
+      expect(fs.writeFile).toHaveBeenCalledTimes(1);
+      expect(fs.writeFile).toHaveBeenCalledWith(
+        sessionPath,
+        expect.stringContaining('Add dark mode to the app'),
       );
     });
 
-    it('should not call saveSummary() if generation fails', async () => {
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        messages: [
-          {
-            id: '1',
-            timestamp: '2025-12-03T00:00:00Z',
-            type: 'user' as const,
-            content: [{ text: 'Hello' }],
-          },
-        ],
-      };
+    it('should handle errors gracefully without throwing', async () => {
+      vi.mocked(fs.access).mockResolvedValue(undefined);
+      mockReaddir.mockResolvedValue(['session-2024-01-01T10-00-abc12345.json']);
+      vi.mocked(fs.readFile).mockResolvedValue(
+        createSessionWithUserMessages(2),
+      );
+      mockGenerateSummary.mockRejectedValue(new Error('API Error'));
 
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
-      mockGenerateSummary.mockResolvedValue(null);
-
-      await generateAndSaveSummary(mockConfig);
-
-      expect(mockGenerateSummary).toHaveBeenCalledTimes(1);
-      expect(mockChatRecordingService.saveSummary).not.toHaveBeenCalled();
-    });
-
-    it('should not call saveSummary() if generateSummary throws', async () => {
-      const mockConversation = {
-        sessionId: 'test-session',
-        projectHash: 'test-hash',
-        startTime: '2025-12-03T00:00:00Z',
-        lastUpdated: '2025-12-03T00:10:00Z',
-        messages: [
-          {
-            id: '1',
-            timestamp: '2025-12-03T00:00:00Z',
-            type: 'user' as const,
-            content: [{ text: 'Hello' }],
-          },
-        ],
-      };
-
-      (
-        mockChatRecordingService.getConversation as ReturnType<typeof vi.fn>
-      ).mockReturnValue(mockConversation);
-      mockGenerateSummary.mockRejectedValue(new Error('Generation failed'));
-
-      await generateAndSaveSummary(mockConfig);
-
-      expect(mockGenerateSummary).toHaveBeenCalledTimes(1);
-      expect(mockChatRecordingService.saveSummary).not.toHaveBeenCalled();
+      await expect(generateSummary(mockConfig)).resolves.not.toThrow();
     });
   });
 });

packages/core/src/services/sessionSummaryUtils.ts

@@ -8,62 +8,172 @@ import type { Config } from '../config/config.js';
 import { SessionSummaryService } from './sessionSummaryService.js';
 import { BaseLlmClient } from '../core/baseLlmClient.js';
 import { debugLogger } from '../utils/debugLogger.js';
+import {
+  SESSION_FILE_PREFIX,
+  type ConversationRecord,
+} from './chatRecordingService.js';
+import fs from 'node:fs/promises';
+import path from 'node:path';
+
+const MIN_MESSAGES_FOR_SUMMARY = 1;
 
 /**
- * Generates and saves a summary for the current session.
- * This is called during session cleanup and is non-blocking - errors are logged but don't prevent exit.
+ * Generates and saves a summary for a session file.
  */
-export async function generateAndSaveSummary(config: Config): Promise<void> {
+async function generateAndSaveSummary(
+  config: Config,
+  sessionPath: string,
+): Promise<void> {
+  // Read session file
+  const content = await fs.readFile(sessionPath, 'utf-8');
+  const conversation: ConversationRecord = JSON.parse(content);
+
+  // Skip if summary already exists
+  if (conversation.summary) {
+    debugLogger.debug(
+      `[SessionSummary] Summary already exists for ${sessionPath}, skipping`,
+    );
+    return;
+  }
+
+  // Skip if no messages
+  if (conversation.messages.length === 0) {
+    debugLogger.debug(
+      `[SessionSummary] No messages to summarize in ${sessionPath}`,
+    );
+    return;
+  }
+
+  // Create summary service
+  const contentGenerator = config.getContentGenerator();
+  if (!contentGenerator) {
+    debugLogger.debug(
+      '[SessionSummary] Content generator not available, skipping summary generation',
+    );
+    return;
+  }
+  const baseLlmClient = new BaseLlmClient(contentGenerator, config);
+  const summaryService = new SessionSummaryService(baseLlmClient);
+
+  // Generate summary
+  const summary = await summaryService.generateSummary({
+    messages: conversation.messages,
+  });
+
+  if (!summary) {
+    debugLogger.warn(
+      `[SessionSummary] Failed to generate summary for ${sessionPath}`,
+    );
+    return;
+  }
+
+  // Re-read the file before writing to handle race conditions
+  const freshContent = await fs.readFile(sessionPath, 'utf-8');
+  const freshConversation: ConversationRecord = JSON.parse(freshContent);
+
+  // Check if summary was added by another process
+  if (freshConversation.summary) {
+    debugLogger.debug(
+      `[SessionSummary] Summary was added by another process for ${sessionPath}`,
+    );
+    return;
+  }
+
+  // Add summary and write back
+  freshConversation.summary = summary;
+  freshConversation.lastUpdated = new Date().toISOString();
+  await fs.writeFile(sessionPath, JSON.stringify(freshConversation, null, 2));
+  debugLogger.debug(
+    `[SessionSummary] Saved summary for ${sessionPath}: "${summary}"`,
+  );
+}
+
+/**
+ * Finds the most recently created session that needs a summary.
+ * Returns the path if it needs a summary, null otherwise.
+ */
+export async function getPreviousSession(
+  config: Config,
+): Promise<string | null> {
   try {
-    // Get the chat recording service from config
-    const chatRecordingService = config
-      .getGeminiClient()
-      ?.getChatRecordingService();
-    if (!chatRecordingService) {
-      debugLogger.debug('[SessionSummary] No chat recording service available');
-      return;
+    const chatsDir = path.join(config.storage.getProjectTempDir(), 'chats');
+
+    // Check if chats directory exists
+    try {
+      await fs.access(chatsDir);
+    } catch {
+      debugLogger.debug('[SessionSummary] No chats directory found');
+      return null;
     }
 
-    // Get the current conversation
-    const conversation = chatRecordingService.getConversation();
-    if (!conversation) {
-      debugLogger.debug('[SessionSummary] No conversation to summarize');
-      return;
+    // List session files
+    const allFiles = await fs.readdir(chatsDir);
+    const sessionFiles = allFiles.filter(
+      (f) => f.startsWith(SESSION_FILE_PREFIX) && f.endsWith('.json'),
+    );
+
+    if (sessionFiles.length === 0) {
+      debugLogger.debug('[SessionSummary] No session files found');
+      return null;
     }
 
-    // Skip if summary already exists (e.g., resumed session)
-    if (conversation.summary) {
-      debugLogger.debug('[SessionSummary] Summary already exists, skipping');
-      return;
+    // Sort by filename descending (most recently created first)
+    // Filename format: session-YYYY-MM-DDTHH-MM-XXXXXXXX.json
+    sessionFiles.sort((a, b) => b.localeCompare(a));
+
+    // Check the most recently created session
+    const mostRecentFile = sessionFiles[0];
+    const filePath = path.join(chatsDir, mostRecentFile);
+
+    try {
+      const content = await fs.readFile(filePath, 'utf-8');
+      const conversation: ConversationRecord = JSON.parse(content);
+
+      if (conversation.summary) {
+        debugLogger.debug(
+          '[SessionSummary] Most recent session already has summary',
+        );
+        return null;
+      }
+
+      // Only generate summaries for sessions with more than 1 user message
+      const userMessageCount = conversation.messages.filter(
+        (m) => m.type === 'user',
+      ).length;
+      if (userMessageCount <= MIN_MESSAGES_FOR_SUMMARY) {
+        debugLogger.debug(
+          `[SessionSummary] Most recent session has ${userMessageCount} user message(s), skipping (need more than ${MIN_MESSAGES_FOR_SUMMARY})`,
+        );
+        return null;
+      }
+
+      return filePath;
+    } catch {
+      debugLogger.debug('[SessionSummary] Could not read most recent session');
+      return null;
     }
+  } catch (error) {
+    debugLogger.debug(
+      `[SessionSummary] Error finding previous session: ${error instanceof Error ? error.message : String(error)}`,
+    );
+    return null;
+  }
+}
 
-    // Skip if no messages
-    if (conversation.messages.length === 0) {
-      debugLogger.debug('[SessionSummary] No messages to summarize');
-      return;
-    }
-
-    // Create summary service
-    const contentGenerator = config.getContentGenerator();
-    const baseLlmClient = new BaseLlmClient(contentGenerator, config);
-    const summaryService = new SessionSummaryService(baseLlmClient);
-
-    // Generate summary
-    const summary = await summaryService.generateSummary({
-      messages: conversation.messages,
-    });
-
-    // Save summary if generated successfully
-    if (summary) {
-      chatRecordingService.saveSummary(summary);
-      debugLogger.debug(`[SessionSummary] Saved summary: "${summary}"`);
-    } else {
-      debugLogger.warn('[SessionSummary] Failed to generate summary');
+/**
+ * Generates summary for the previous session if it lacks one.
+ * This is designed to be called fire-and-forget on startup.
+ */
+export async function generateSummary(config: Config): Promise<void> {
+  try {
+    const sessionPath = await getPreviousSession(config);
+    if (sessionPath) {
+      await generateAndSaveSummary(config, sessionPath);
     }
   } catch (error) {
     // Log but don't throw - we want graceful degradation
     debugLogger.warn(
-      `[SessionSummary] Error in generateAndSaveSummary: ${error instanceof Error ? error.message : String(error)}`,
+      `[SessionSummary] Error generating summary: ${error instanceof Error ? error.message : String(error)}`,
     );
   }
 }

packages/test-utils/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@google/gemini-cli-test-utils",
-  "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+  "version": "0.21.0-preview.4",
   "private": true,
   "main": "src/index.ts",
   "license": "Apache-2.0",

packages/vscode-ide-companion/package.json

@@ -2,7 +2,7 @@
   "name": "gemini-cli-vscode-ide-companion",
   "displayName": "Gemini CLI Companion",
   "description": "Enable Gemini CLI with direct access to your IDE workspace.",
-  "version": "0.21.0-nightly.20251209.ec9a8c7a7",
+  "version": "0.21.0-preview.4",
   "publisher": "google",
   "icon": "assets/icon.png",
   "repository": {