codex-session-insights 0.1.0 → 0.2.0

package/README.md

@@ -108,8 +108,8 @@ npx codex-session-insights --provider openai --api-key $OPENAI_API_KEY
 Current default analysis plan:
 
 - `days`: `30`
-- `limit`: `50`
-- `facet-limit`: `20`
+- `limit`: `200`
+- `facet-limit`: `50`
 - `provider`: `codex-cli`
 - `facet-model`: `gpt-5.4-mini`
 - `fast-section-model`: `gpt-5.4-mini`
@@ -120,6 +120,8 @@ Current default analysis plan:
 
 Important behavior defaults:
 
+- `limit` means the target number of substantive threads to include in the report, not just the first 50 indexed threads
+- `facet-limit` means the max number of uncached per-thread facet analyses to run in a single report
 - Report language follows a best-effort system locale check
 - Main-thread analysis is the default; sub-agent threads are excluded unless you pass `--include-subagents`
 - The CLI shows an estimate before running in interactive terminals

package/lib/cli.js

@@ -141,7 +141,7 @@ function parseArgs(argv) {
     jsonPath: null,
     htmlPath: null,
     days: 30,
-    limit: 50,
+    limit: 200,
     preview: 50,
     provider: 'codex-cli',
     codexBin: null,
@@ -154,7 +154,7 @@ function parseArgs(argv) {
     insightModel: null,
     insightEffort: null,
     cacheDir: null,
-    facetLimit: 20,
+    facetLimit: 50,
     lang: detectSystemLanguage(),
     includeArchived: false,
     includeSubagents: false,
@@ -330,7 +330,7 @@ Options:
   --json-path <path>        Exact path for report.json
   --html-path <path>        Exact path for report.html
   --days <n>                Only include threads updated in the last N days (default: 30)
-  --limit <n>               Max threads to scan from the thread registry (default: 50)
+  --limit <n>               Target number of substantive threads to include (default: 200)
   --preview <n>             Number of threads to embed in the HTML report (default: 50)
   --provider <name>         Model provider: codex-cli or openai (default: codex-cli)
   --codex-bin <path>        Override the Codex CLI binary path for provider=codex-cli
@@ -344,7 +344,7 @@ Options:
                             Reasoning effort for lower-risk sections
   --insight-model <name>    Model for final report generation
   --insight-effort <level>  Reasoning effort for higher-risk sections
-  --facet-limit <n>         Max uncached substantive threads to analyze (default: 20)
+  --facet-limit <n>         Max uncached thread facets to analyze (default: 50)
   --cache-dir <path>        Cache directory for session-meta and facet caches
   --lang <code>             Report language: en or zh-CN (default: system language)
   --include-archived        Include archived threads
@@ -657,8 +657,8 @@ function inferScopePreset(days) {
 
 function inferDepthPreset(options) {
   if (options.limit === 20 && options.facetLimit === 8) return 'conservative'
-  if (options.limit === 50 && options.facetLimit === 20) return 'standard'
-  if (options.limit === 200 && options.facetLimit === 50) return 'deep'
+  if (options.limit === 200 && options.facetLimit === 50) return 'standard'
+  if (options.limit === 400 && options.facetLimit === 50) return 'deep'
   return 'custom'
 }
 
@@ -682,8 +682,8 @@ function inferQualityPreset(options) {
 
 function applyScopePreset(options, preset) {
   if (preset === 'conservative') return { ...options, limit: 20, facetLimit: 8 }
-  if (preset === 'deep') return { ...options, limit: 200, facetLimit: 50 }
-  return { ...options, limit: 50, facetLimit: 20 }
+  if (preset === 'deep') return { ...options, limit: 400, facetLimit: 50 }
+  return { ...options, limit: 200, facetLimit: 50 }
 }
 
 function applyQualityPreset(options, preset) {
@@ -833,17 +833,17 @@ function getUiText(lang) {
       scopeCustom: '自定义天数',
       customDaysQuestion: '输入要分析的天数',
       depthQuestion: '选择分析深度：',
-      depthConservative: '保守（20 threads / 8 facets）',
-      depthStandard: '标准（50 threads / 20 facets）',
-      depthDeep: '深度（200 threads / 50 facets）',
+      depthConservative: '保守（20 个有效线程 / 8 个 facets）',
+      depthStandard: '标准（200 个有效线程 / 50 个 facets）',
+      depthDeep: '深度（400 个有效线程 / 50 个 facets）',
       depthCustom: '自定义',
-      limitQuestion: '输入最大 thread 数',
-      facetLimitQuestion: '输入最大 facet 提取数',
+      limitQuestion: '输入目标有效线程数',
+      facetLimitQuestion: '输入最大新增 facet 数',
       languageQuestion: '选择报告语言：',
       outputDirQuestion: '输出目录',
       openBrowserQuestion: '生成后自动打开浏览器？',
       qualityQuestion: '选择分析质量预设：',
-      qualityCheaper: '更省（尽量用 spark）',
+      qualityCheaper: '更省（尽量用 gpt-5.4-mini）',
       qualityBalanced: '平衡',
       qualityHigher: '更高质量（更多使用 gpt-5.4）',
       yesDefault: '[Y/n]',
@@ -860,7 +860,7 @@ function getUiText(lang) {
       toWord: '到',
       likelyWord: '左右',
       plannedCallsLabel: '预计调用数',
-      substantiveThreadsLabel: '有效线程',
+      substantiveThreadsLabel: '纳入报告线程',
       uncachedFacetsLabel: '未缓存 facets',
       longTranscriptsLabel: '长 transcript',
       inputEstimateLabel: '输入',
@@ -891,17 +891,17 @@ function getUiText(lang) {
     scopeCustom: 'Custom days',
     customDaysQuestion: 'Enter number of days to analyze',
     depthQuestion: 'Choose analysis depth:',
-    depthConservative: 'Conservative (20 threads / 8 facets)',
-    depthStandard: 'Standard (50 threads / 20 facets)',
-    depthDeep: 'Deep (200 threads / 50 facets)',
+    depthConservative: 'Conservative (20 substantive threads / 8 facets)',
+    depthStandard: 'Standard (200 substantive threads / 50 facets)',
+    depthDeep: 'Deep (400 substantive threads / 50 facets)',
     depthCustom: 'Custom',
-    limitQuestion: 'Enter max thread count',
-    facetLimitQuestion: 'Enter max facet extraction count',
+    limitQuestion: 'Enter target substantive thread count',
+    facetLimitQuestion: 'Enter max new facet extraction count',
     languageQuestion: 'Choose report language:',
     outputDirQuestion: 'Output directory',
     openBrowserQuestion: 'Open the report in your browser after generation?',
     qualityQuestion: 'Choose quality preset:',
-    qualityCheaper: 'Cheaper (more spark)',
+    qualityCheaper: 'Cheaper (more gpt-5.4-mini)',
     qualityBalanced: 'Balanced',
     qualityHigher: 'Higher quality (more gpt-5.4)',
     yesDefault: '[Y/n]',
@@ -918,7 +918,7 @@ function getUiText(lang) {
     toWord: 'to',
     likelyWord: 'likely',
     plannedCallsLabel: 'planned calls',
-    substantiveThreadsLabel: 'substantive threads',
+    substantiveThreadsLabel: 'threads in report',
     uncachedFacetsLabel: 'uncached facets',
     longTranscriptsLabel: 'long transcripts',
     inputEstimateLabel: 'input',

package/lib/codex-data.js

@@ -44,6 +44,7 @@ export async function loadThreads({
   codexHome,
   sinceEpochSeconds,
   limit,
+  offset,
   includeArchived,
   includeSubagents,
 }) {
@@ -66,6 +67,7 @@ export async function loadThreads({
     where.length ? `where ${where.join(' and ')}` : '',
     'order by updated_at desc',
     limit ? `limit ${Number(limit)}` : '',
+    offset ? `offset ${Number(offset)}` : '',
     ';',
   ]
     .filter(Boolean)
@@ -115,11 +117,15 @@ export async function loadRolloutEvents(rolloutPath) {
   return events
 }
 
+export function isSubstantiveThread(thread) {
+  if (thread.userMessages < 2) return false
+  if (thread.durationMinutes < 1) return false
+  return Boolean(String(thread.transcriptForAnalysis || '').trim())
+}
+
 export function filterSubstantiveThreads(threadSummaries) {
   return [...threadSummaries]
-    .filter(thread => thread.userMessages >= 2)
-    .filter(thread => thread.durationMinutes >= 1)
-    .filter(thread => Boolean(String(thread.transcriptForAnalysis || '').trim()))
+    .filter(isSubstantiveThread)
     .sort((a, b) => Date.parse(b.updatedAt) - Date.parse(a.updatedAt))
 }
 
@@ -440,14 +446,43 @@ export function summarizeThread(thread, events) {
 }
 
 export async function collectThreadSummaries(options) {
-  const threads = await loadThreads(options)
   const cacheDir = resolveSessionMetaCacheDir(options.cacheDir)
   await fs.mkdir(cacheDir, { recursive: true })
 
+  const substantiveTarget = Number(options.limit ?? 0)
+  const pageSize = substantiveTarget > 0 ? substantiveTarget : 200
   const summaries = []
-  for (const thread of threads) {
-    const summary = await loadCachedOrFreshSummary(thread, cacheDir)
-    summaries.push(summary)
+  let offset = 0
+  let substantiveCount = 0
+
+  while (true) {
+    const threads = await loadThreads({
+      ...options,
+      limit: pageSize,
+      offset,
+    })
+    if (!threads.length) break
+
+    for (const thread of threads) {
+      const summary = await loadCachedOrFreshSummary(thread, cacheDir)
+      summaries.push(summary)
+      if (isSubstantiveThread(summary)) {
+        substantiveCount += 1
+      }
+    }
+
+    if (substantiveTarget > 0 && substantiveCount >= substantiveTarget) {
+      break
+    }
+    if (threads.length < pageSize) {
+      break
+    }
+
+    offset += threads.length
+  }
+
+  if (substantiveTarget > 0) {
+    return filterSubstantiveThreads(summaries).slice(0, substantiveTarget)
   }
   return summaries
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "codex-session-insights",
-  "version": "0.1.0",
+  "version": "0.2.0",
   "description": "Generate a report analyzing your Codex sessions.",
   "type": "module",
   "bin": {