Export Script Deep Dive

The export script (scripts/export.mjs) is a standalone Node.js program that reads your local ~/.claude/ directory and generates a single JSON file containing all your usage data.

The script uses only Node.js built-in modules (fs, path, os). No external dependencies required.

Usage

node scripts/export.mjs

Output file: claude-analytics-export.json (in current directory)

What Gets Collected

The script collects five types of data:

1. Stats Cache

Source: ~/.claude/stats-cache.json Contains:

Total sessions and messages
Daily activity (messages, sessions, tool calls per day)
Model usage (tokens, cache hits, costs per model)
Hourly activity distribution
Longest session metadata

Code:

// From: scripts/export.mjs (lines 12-19)
let stats = null;
try {
  stats = JSON.parse(fs.readFileSync(path.join(CLAUDE_DIR, "stats-cache.json"), "utf-8"));
  console.log("  stats-cache.json ✓");
} catch {
  console.log("  stats-cache.json ✗ (not found)");
}

If the stats cache doesn’t exist, the script continues gracefully with stats = null.

2. Session Metadata

Source: ~/.claude/usage-data/session-meta/*.json Contains (per session):

Session ID and project path
Start time and duration
Message counts (user + assistant)
Tool usage counts by type
Programming languages used
Git activity (commits, pushes)
Token counts (input, output)
Code changes (lines added/removed, files modified)

Code:

// From: scripts/export.mjs (lines 21-37)
let sessions = [];
const metaDir = path.join(CLAUDE_DIR, "usage-data", "session-meta");
try {
  const files = fs.readdirSync(metaDir).filter((f) => f.endsWith(".json"));
  for (const f of files) {
    try {
      const raw = fs.readFileSync(path.join(metaDir, f), "utf-8");
      const meta = JSON.parse(raw);
      if (meta.duration_minutes > 0) sessions.push(meta);
    } catch { /* skip */ }
  }
  sessions.sort((a, b) => new Date(b.start_time).getTime() - new Date(a.start_time).getTime());
  console.log(`  session-meta/ ✓ (${sessions.length} sessions)`);
} catch {
  console.log("  session-meta/ ✗ (not found)");
}

Sessions with duration_minutes = 0 are filtered out as they represent aborted or invalid sessions.

3. Prompt History

Source: ~/.claude/history.jsonl Format: JSONL (one JSON object per line) Contains (per entry):

display: The prompt text
timestamp: Unix timestamp (milliseconds)
project: Project identifier
pastedContents: Attached file/context metadata

Code:

// From: scripts/export.mjs (lines 39-49)
let history = [];
try {
  const raw = fs.readFileSync(path.join(CLAUDE_DIR, "history.jsonl"), "utf-8");
  history = raw.trim().split("\n").map((line) => {
    try { return JSON.parse(line); } catch { return null; }
  }).filter(Boolean);
  console.log(`  history.jsonl ✓ (${history.length} entries)`);
} catch {
  console.log("  history.jsonl ✗ (not found)");
}

Invalid lines are skipped rather than causing the entire export to fail.

4. Account Information

Source: ~/.claude/statsig/*cached.evaluations*.json Contains:

Account UUID
Organization UUID (if applicable)

Purpose: Enables multi-profile support in hosted mode Code:

// From: scripts/export.mjs (lines 51-71)
let account = {};
try {
  const statsigDir = path.join(CLAUDE_DIR, "statsig");
  const files = fs.readdirSync(statsigDir).filter((f) => f.includes("cached.evaluations"));
  for (const f of files) {
    try {
      const raw = fs.readFileSync(path.join(statsigDir, f), "utf-8");
      const parsed = JSON.parse(raw);
      if (parsed.user?.userID) account.accountUUID = parsed.user.userID;
      if (parsed.user?.custom?.organization_uuid) account.organizationUUID = parsed.user.custom.organization_uuid;
    } catch { /* skip */ }
  }
  if (account.accountUUID) {
    console.log(`  account info ✓ (${account.accountUUID.slice(0, 8)}...)`);
  } else {
    console.log("  account info ✗ (not found in statsig cache)");
  }
} catch {
  console.log("  account info ✗ (statsig dir not found)");
}

Account info is optional. The export succeeds even if this data is missing.

5. Project Memories

Source: ~/.claude/projects/*/memory/*.md Contains: Markdown files that Claude Code uses to remember project context Code:

// From: scripts/export.mjs (lines 73-99)
let memories = [];
try {
  const projectsDir = path.join(CLAUDE_DIR, "projects");
  const projects = fs.readdirSync(projectsDir);
  for (const p of projects) {
    const memDir = path.join(projectsDir, p, "memory");
    try {
      const mdFiles = fs.readdirSync(memDir).filter((f) => f.endsWith(".md"));
      if (mdFiles.length > 0) {
        const files = mdFiles.map((f) => ({
          name: f,
          content: fs.readFileSync(path.join(memDir, f), "utf-8").slice(0, 5000),
        }));
        memories.push({ project: p, files });
      }
    } catch { /* no memory dir */ }
  }
  if (memories.length > 0) {
    const totalFiles = memories.reduce((a, m) => a + m.files.length, 0);
    console.log(`  memories ✓ (${totalFiles} files across ${memories.length} projects)`);
  } else {
    console.log("  memories ✗ (none found)");
  }
} catch {
  console.log("  memories ✗ (projects dir not found)");
}

Each memory file is truncated to 5,000 characters to prevent massive exports. If you need full content, use local mode instead.

Output Format

The script generates a single JSON file with this structure:

interface ExportData {
  stats: StatsCache | null;
  sessions: SessionMeta[];
  history: HistoryEntry[];
  memories: ProjectMemory[];
  account?: {
    accountUUID?: string;
    organizationUUID?: string;
  };
  exportedAt: string;  // ISO 8601 timestamp
}

Writing the Output

// From: scripts/export.mjs (lines 101-114)
const data = {
  stats,
  sessions,
  history,
  memories,
  account: Object.keys(account).length > 0 ? account : undefined,
  exportedAt: new Date().toISOString(),
};
fs.writeFileSync(OUTPUT, JSON.stringify(data));

const sizeMB = (fs.statSync(OUTPUT).size / 1024 / 1024).toFixed(1);
console.log(`\nExported to: ${OUTPUT} (${sizeMB} MB)`);
console.log("Upload this file at: https://claude-analytics.vercel.app");

Example Output

Claude Code Analytics — Data Export

  stats-cache.json ✓
  session-meta/ ✓ (127 sessions)
  history.jsonl ✓ (1,543 entries)
  account info ✓ (a1b2c3d4...)
  memories ✓ (23 files across 5 projects)

Exported to: /Users/you/claude-analytics-export.json (2.4 MB)
Upload this file at: https://claude-analytics.vercel.app

File Size Breakdown

Typical export sizes:

Component	Lines/Files	Size
Stats cache	1 file	~50 KB
Session metadata	100 sessions	~200 KB
Prompt history	1,000 entries	~500 KB
Project memories	20 files	~100 KB
Total		~850 KB

Exports scale linearly with usage. Heavy users (5,000+ prompts) may see 5-10 MB files.

Customizing the Export

Filter by Date Range

Only export sessions from the last 30 days:

// After loading sessions (line 37)
const thirtyDaysAgo = Date.now() - 30 * 24 * 60 * 60 * 1000;
sessions = sessions.filter(s => new Date(s.start_time).getTime() > thirtyDaysAgo);

Exclude Prompt History

Comment out lines 39-49:

// let history = [];
// try { ... } catch { ... }
history = [];  // Export empty array instead

Include Full Memory Files

Remove the .slice(0, 5000) truncation (line 85):

content: fs.readFileSync(path.join(memDir, f), "utf-8"),  // Full content

Removing truncation can result in very large export files (> 50 MB) if you have extensive project memories.

Add Custom Metadata

Extend the output object (line 102):

const data = {
  stats,
  sessions,
  history,
  memories,
  account: Object.keys(account).length > 0 ? account : undefined,
  exportedAt: new Date().toISOString(),
  // Custom fields:
  exportVersion: "2.0",
  machineName: os.hostname(),
  nodeVersion: process.version,
};

Data Privacy

The export script:

✓ Runs entirely locally (no network requests)
✓ Only reads from ~/.claude/ (no other directories)
✓ Uses built-in Node.js modules (no third-party dependencies)
✓ Writes to current directory (no system-wide changes)
✓ Is fully open source and auditable

Review the script yourself: It’s only 115 lines of readable JavaScript with zero dependencies.

Dependencies

Zero external dependencies. The script only imports Node.js built-ins:

import fs from "fs";
import path from "path";
import { homedir } from "os";

From package.json:

{
  "scripts": {
    "export-data": "node scripts/export.mjs"
  }
}

You can run it with:

node scripts/export.mjs (Node.js 18+)
bun scripts/export.mjs (Bun)
deno run --allow-read --allow-write scripts/export.mjs (Deno)

Troubleshooting

Error: Cannot find module 'fs'

You’re using an outdated Node version. Update to Node.js 18+:

node --version
# If < 18, update via: https://nodejs.org

All data sources show ✗ (not found)

Your ~/.claude/ directory doesn’t exist or is empty. Verify:

ls -la ~/.claude/

If empty, use Claude Code at least once to generate data.

Export file is only 2 bytes

The script ran but found no data. Check the console output to see which sources failed.Minimal valid export:

{"stats":null,"sessions":[],"history":[],"memories":[]}

Permission denied writing output file

Ensure you have write access to the current directory:

pwd
ls -la .

Or specify a different output location:

// Edit line 8 in export.mjs:
const OUTPUT = "/tmp/claude-export.json";

Script hangs or takes a very long time

You may have an extremely large history.jsonl file. Check its size:

ls -lh ~/.claude/history.jsonl

If > 50 MB, consider filtering to recent entries only (see customization above).

Complete Script Source

Here’s the full export script for reference:

#!/usr/bin/env node

import fs from "fs";
import path from "path";
import { homedir } from "os";

const CLAUDE_DIR = path.join(homedir(), ".claude");
const OUTPUT = path.join(process.cwd(), "claude-analytics-export.json");

console.log("Claude Code Analytics — Data Export\n");

// 1. Stats cache
let stats = null;
try {
  stats = JSON.parse(fs.readFileSync(path.join(CLAUDE_DIR, "stats-cache.json"), "utf-8"));
  console.log("  stats-cache.json ✓");
} catch {
  console.log("  stats-cache.json ✗ (not found)");
}

// 2. Session metas
let sessions = [];
const metaDir = path.join(CLAUDE_DIR, "usage-data", "session-meta");
try {
  const files = fs.readdirSync(metaDir).filter((f) => f.endsWith(".json"));
  for (const f of files) {
    try {
      const raw = fs.readFileSync(path.join(metaDir, f), "utf-8");
      const meta = JSON.parse(raw);
      if (meta.duration_minutes > 0) sessions.push(meta);
    } catch { /* skip */ }
  }
  sessions.sort((a, b) => new Date(b.start_time).getTime() - new Date(a.start_time).getTime());
  console.log(`  session-meta/ ✓ (${sessions.length} sessions)`);
} catch {
  console.log("  session-meta/ ✗ (not found)");
}

// 3. History
let history = [];
try {
  const raw = fs.readFileSync(path.join(CLAUDE_DIR, "history.jsonl"), "utf-0");
  history = raw.trim().split("\n").map((line) => {
    try { return JSON.parse(line); } catch { return null; }
  }).filter(Boolean);
  console.log(`  history.jsonl ✓ (${history.length} entries)`);
} catch {
  console.log("  history.jsonl ✗ (not found)");
}

// 4. Account info
let account = {};
try {
  const statsigDir = path.join(CLAUDE_DIR, "statsig");
  const files = fs.readdirSync(statsigDir).filter((f) => f.includes("cached.evaluations"));
  for (const f of files) {
    try {
      const raw = fs.readFileSync(path.join(statsigDir, f), "utf-8");
      const parsed = JSON.parse(raw);
      if (parsed.user?.userID) account.accountUUID = parsed.user.userID;
      if (parsed.user?.custom?.organization_uuid) account.organizationUUID = parsed.user.custom.organization_uuid;
    } catch { /* skip */ }
  }
  if (account.accountUUID) {
    console.log(`  account info ✓ (${account.accountUUID.slice(0, 8)}...)`);
  } else {
    console.log("  account info ✗ (not found in statsig cache)");
  }
} catch {
  console.log("  account info ✗ (statsig dir not found)");
}

// 5. Project memories (with file content)
let memories = [];
try {
  const projectsDir = path.join(CLAUDE_DIR, "projects");
  const projects = fs.readdirSync(projectsDir);
  for (const p of projects) {
    const memDir = path.join(projectsDir, p, "memory");
    try {
      const mdFiles = fs.readdirSync(memDir).filter((f) => f.endsWith(".md"));
      if (mdFiles.length > 0) {
        const files = mdFiles.map((f) => ({
          name: f,
          content: fs.readFileSync(path.join(memDir, f), "utf-8").slice(0, 5000),
        }));
        memories.push({ project: p, files });
      }
    } catch { /* no memory dir */ }
  }
  if (memories.length > 0) {
    const totalFiles = memories.reduce((a, m) => a + m.files.length, 0);
    console.log(`  memories ✓ (${totalFiles} files across ${memories.length} projects)`);
  } else {
    console.log("  memories ✗ (none found)");
  }
} catch {
  console.log("  memories ✗ (projects dir not found)");
}

// Write output
const data = {
  stats,
  sessions,
  history,
  memories,
  account: Object.keys(account).length > 0 ? account : undefined,
  exportedAt: new Date().toISOString(),
};
fs.writeFileSync(OUTPUT, JSON.stringify(data));

const sizeMB = (fs.statSync(OUTPUT).size / 1024 / 1024).toFixed(1);
console.log(`\nExported to: ${OUTPUT} (${sizeMB} MB)`);
console.log("Upload this file at: https://claude-analytics.vercel.app");

Next Steps

Upload to Hosted Mode

Learn how to upload your export to the hosted app

Understanding Data

Deep dive into what each data source contains

Data Model Reference

Complete TypeScript interfaces for all data types

Privacy & Security

Learn about data security guarantees

Get Started

Setup

Features

Guides

Usage

What Gets Collected

1. Stats Cache

2. Session Metadata

3. Prompt History

4. Account Information

5. Project Memories

Output Format

Writing the Output

Example Output

File Size Breakdown

Customizing the Export

Filter by Date Range

Exclude Prompt History

Include Full Memory Files

Add Custom Metadata

Data Privacy

Dependencies

Troubleshooting

Complete Script Source

Next Steps

Upload to Hosted Mode

Understanding Data

Data Model Reference

Privacy & Security

Build docs developers (and LLMs) love

Get Started

Setup

Features

Guides

​Usage

​What Gets Collected

​1. Stats Cache

​2. Session Metadata

​3. Prompt History

​4. Account Information

​5. Project Memories

​Output Format

​Writing the Output

​Example Output

​File Size Breakdown

​Customizing the Export

​Filter by Date Range

​Exclude Prompt History

​Include Full Memory Files

​Add Custom Metadata

​Data Privacy

​Dependencies

​Troubleshooting

​Complete Script Source

​Next Steps

Upload to Hosted Mode

Understanding Data

Data Model Reference

Privacy & Security

Build docs developers (and LLMs) love

Usage

What Gets Collected

1. Stats Cache

2. Session Metadata

3. Prompt History

4. Account Information

5. Project Memories

Output Format

Writing the Output

Example Output

File Size Breakdown

Customizing the Export

Filter by Date Range

Exclude Prompt History

Include Full Memory Files

Add Custom Metadata

Data Privacy

Dependencies

Troubleshooting

Complete Script Source

Next Steps