Discord Data Explorer

Offline CLI tool (ddex) that analyses a Discord GDPR data export and reports the words you type most often. Similar-sounding words (e.g. haha, hahah, hahsha) are automatically clustered together.

All processing is entirely local, no network calls are made.

Download

Pre-built standalone binaries are available on the releases page. No runtime or dependencies required: just download and run.

Releases are created automatically when a new version is merged. Each release includes a CHANGELOG.md entry describing what changed.

Platform	File
Linux x64	ddex-linux-x64
Linux ARM64	ddex-linux-arm64
macOS Intel	ddex-macos-x64
macOS Apple Silicon	ddex-macos-arm64
Windows x64	ddex-windows-x64.exe

On macOS/Linux you may need to mark the binary as executable:

chmod +x ddex-*

Getting your Discord data package

1. Open Discord → User Settings → Data & Privacy → Request Your Data
2. Discord emails you a download link (usually within 30 days)
3. Download the .zip: pass it directly to ddex

Your files are never modified. If a .zip is provided it is extracted to a temporary directory which is deleted immediately after processing.

Usage

ddex <command> [options]

Commands:
  words        Analyse your most-used words in a Discord export
  prediction   Show Discord's predicted age group and gender
  spent        Show how much you have spent on Discord
  people       Show your social graph and DM statistics
  servers      Show your server and channel activity
  time         Show your temporal activity patterns
  emojis       Show your emoji usage and reactions
  attachments  Show your attachment activity
  stats        Show a full summary dashboard of your Discord activity

Options:
  --version, -V   Print version and exit
  --help, -h      Show this help message

ddex words

ddex words <path-to-export> [options]

Arguments:
  path-to-export          Path to your Discord data package.
                          Accepts a directory (unzipped export) or a .zip file.

Options:
  --top <n>               Show top N word groups (default: 10)
  --include-stop-words    Include stop words in results (filtered by default)
  --language <codes>      Comma-separated stop-word language codes (default: eng)
                          Supports 62 languages, e.g. eng, swe, deu, fra
  --output <file>         Write results to a file (.json for JSON, anything else for plain text)
  --help, -h              Show this help message

Examples

# Show top 20 words
ddex words ~/Downloads/package.zip --top 20

# Pass an already-unzipped export directory
ddex words ~/Downloads/package/ --top 20

# Save results to a JSON file
ddex words ~/Downloads/package.zip --output results.json

# Filter stop words for both English and Swedish
ddex words ~/Downloads/package.zip --language eng,swe

# Include stop words (the, a, is, …) in the results
ddex words ~/Downloads/package.zip --include-stop-words

ddex prediction

ddex prediction <path-to-export>

Arguments:
  path-to-export   Path to your Discord data package.
                   Accepts a directory (unzipped export) or a .zip file.

Options:
  --help, -h       Show this help message

Example output:

Discord's demographic predictions
──────────────────────────────────
Age group:  25-34      (72.3% confidence)
Gender:     male       (89.2% confidence)

Example usage:

ddex prediction ~/Downloads/package.zip
ddex prediction ~/Downloads/package/

ddex spent

Show how much you have spent on Discord (Nitro, gifts, store purchases).

ddex spent ~/Downloads/package.zip

ddex people

Show your social graph: friend count, DM partner count, top DM contacts, and voice call statistics.

ddex people ~/Downloads/package.zip

ddex servers

Show server and channel activity: top servers and text channels by message count, top voice channels by time.

ddex servers ~/Downloads/package.zip

ddex time

Show temporal activity patterns: message heatmap by day/hour, monthly activity, session stats and OS breakdown.

ddex time ~/Downloads/package.zip

ddex emojis

Show your most-used emojis in messages and reactions given (from analytics).

ddex emojis ~/Downloads/package.zip

ddex attachments

Show attachment activity: total attachments sent and top channels by attachment count.

ddex attachments ~/Downloads/package.zip

ddex stats

Show a full summary dashboard aggregating highlights from all other commands in a single pass.

ddex stats ~/Downloads/package.zip

---

Development

Requirements

• Bun ≥ 1.3

Setup

git clone https://github.com/Osyx/ddex.git
cd ddex
bun install

Building

For your current platform:

bun run build

Cross-platform builds (output goes to dist/):

bun run build:linux-x64      # Linux x64
bun run build:linux-arm64    # Linux ARM64 (e.g. Raspberry Pi, AWS Graviton)
bun run build:macos-x64      # macOS Intel
bun run build:macos-arm64    # macOS Apple Silicon (M1/M2/M3)
bun run build:windows-x64    # Windows x64 (.exe)
bun run build:all            # All of the above

The resulting binaries are fully standalone.

Scripts

Script	Description
bun run ci	Run all checks and tests (format, lint, unit, integration)
bun run test	Run unit tests
bun run test:integration	Build binary and run integration tests
bun run check	Lint + type-check with oxlint
bun run format	Format source files with oxfmt
bun run format:check	Check formatting without writing changes
bun run build	Compile a binary for the current platform
bun run build:linux-x64	Cross-compile for Linux x64 → dist/
bun run build:linux-arm64	Cross-compile for Linux ARM64 → dist/
bun run build:macos-x64	Cross-compile for macOS Intel → dist/
bun run build:macos-arm64	Cross-compile for macOS Apple Silicon → dist/
bun run build:windows-x64	Cross-compile for Windows x64 → dist/
bun run build:all	Build for all platforms → dist/

How it works

1. Extract: if the input is a .zip, it is extracted to a temp directory
2. Parse: every messages/c*/messages.json file in the export is parsed
3. Tokenise: Discord mentions, URLs, emoji, and markdown are stripped; text is lowercased and split into words (≥ 2 chars, non-numeric)
4. Count: word frequencies are accumulated in an in-memory SQLite database, so memory usage stays flat regardless of message history size
5. Filter: stop words are removed using the stopword package (62 languages supported, default: eng)
6. Cluster: words are grouped by Double Metaphone phonetic key, then further merged by Levenshtein edit distance (threshold: max(1, floor(len/3)))
7. Output: groups are ranked by total count; the most frequent variant is used as the display name

Project structure

src/
  index.ts        CLI entry point, argument parsing, and command dispatch
  cmd.ts          Shared CLI utilities
  extractor.ts    ZIP detection and extraction to temp dir
  parser.ts       Walk export directory, parse messages.json files
  tokenizer.ts    Strip noise, split into word tokens
  stopwords.ts    Stop-word filtering via the stopword package (62 languages)
  db.ts           SQLite-backed word frequency store (bun:sqlite)
  grouper.ts      Phonetic + edit-distance fuzzy clustering
  formatter.ts    Console and file output formatting
  progress.ts     TTY-aware progress reporter (writes to stderr)
  predictor.ts    Locate and stream the analytics events file; extract demographic predictions
  analyze.ts      Single-pass multi-analyzer engine for message files
  analytics.ts    Analytics JSONL streaming and collector dispatch
  metadata.ts     Load user.json, channel.json, and index files
  spent.ts        Discord spending summary from payment records
  people.ts       Social graph and DM partner statistics
  servers.ts      Server and channel activity statistics
  time.ts         Temporal activity patterns (heatmap, session stats)
  emojis.ts       Emoji usage in messages and reactions from analytics
  attachments.ts  Attachment activity per channel
  stats.ts        Single-pass summary dashboard aggregating all analyzers
  types.ts        Shared TypeScript interfaces

CI / GitHub Actions

test.yml: Automated tests

Runs bun test (unit tests) on every pull request. Add this as a required status check in your branch protection rules (Settings → Branches → Branch protection rules → Require status checks → test) to enforce that all tests must pass before merging.

binary-test.yml: Binary integration tests

Runs on every pull request using a matrix of three runners (ubuntu-latest, macos-latest, windows-latest). Each runner builds the platform-specific binary and runs bun run test:binary against the test fixture to verify the compiled binary works end-to-end on that OS. Add the binary-test (ubuntu-latest), binary-test (macos-latest), and binary-test (windows-latest) jobs as required status checks to enforce cross-platform correctness before merging.

release.yml: Automated releases

Triggered on every push to main.

1. release-please analyses conventional commits since the last release and creates (or updates) a Release PR that bumps package.json / CHANGELOG.md to the next semantic version.
2. When that Release PR is merged, release-please tags the commit and creates a GitHub Release.
3. The build job then compiles all five platform binaries and uploads them as assets to that release.

no build runs on regular pushes to main.

Version increments follow Conventional Commits:

Commit prefix	Version bump
fix:	patch
feat:	minor
feat!: / BREAKING CHANGE:	major

conventional-commits.yml: Commit linting

Runs on every pull request and direct push to main. Uses commitlint with the @commitlint/config-conventional preset to reject commits that do not follow the conventional commits format.

Common types: feat, fix, docs, refactor, test, build, ci, chore, perf, revert.

License

This project is licensed under the GNU General Public License v3.0 (GPL-3.0).

You are free to use, share, and modify this software, provided that:

• Attribution: you give appropriate credit and link back to this project
• ShareAlike: any derivative work is distributed under the same licence (GPL-3.0 or later)
• Source disclosure: if you distribute the software, you must make the full source code available