discord-archive

Archive, search, and browse Discord server messages with a fast web viewer.

• Full archive: channels, threads, attachments, reactions, embeds, users
• Incremental updates: only fetches new messages on re-runs
• Full-text search: FTS5-powered search with #channel and @user filters
• Keyboard-driven: use-kbd omnibar (Cmd+K), shortcuts, arrow-key navigation
• Deployable: Cloudflare Workers + D1 + Pages, with GitHub Actions CI/CD
• Versioned data: DVX-tracked archive with S3 remote cache

Quick start

# 1. Set your Discord bot token and guild ID
export DISCORD_TOKEN="your-bot-token"
export DISCORD_GUILD="your-guild-id"

# 2. Archive all messages
./archive.py

# 3. Build the SQLite database
./build_db.py

# 4. Start the local API server
./server.py &

# 5. Start the viewer
cd app && pnpm install && pnpm dev
# Open http://localhost:5272

Architecture

discord-archive/
  archive.py          # Discord API → JSON (incremental, per-channel files)
  build_db.py         # JSON → SQLite (normalized, FTS5 search index)
  build_index.py      # JSON → index.json (for static viewer)
  server.py           # Local dev API server (Starlette + SQLite)
  archive/            # DVX-tracked raw JSON archive + attachments
  archive.db          # SQLite database (derived from archive/)
  app/                # Vite + React viewer
  api/                # Cloudflare Worker (D1-backed API)
    d1-import.sh      # Full SQLite → D1 import
    d1-sync.py        # Incremental D1 sync (zero downtime)
  actions/            # Composite GH Actions for downstream repos
  .github/workflows/  # CI (typecheck)

Scripts

archive.py

Archives all messages from a Discord guild to per-channel JSON files.

./archive.py                          # archive all channels + threads
./archive.py --no-threads             # skip thread messages
./archive.py --no-attachments         # skip downloading attachments
./archive.py --backfill-attachments   # re-fetch expired CDN URLs and download
./archive.py -g 123456789             # specify guild ID
./archive.py -o my-archive            # custom output directory

Requires DISCORD_TOKEN env var (bot token with Message Content intent).

build_db.py

Builds a normalized SQLite database from the JSON archive.

./build_db.py                         # default: archive/ → archive.db
./build_db.py -i my-archive -o my.db  # custom paths

Creates tables: channels, messages, users, attachments, reactions, embeds, threads, plus a messages_fts FTS5 index.

server.py

Local development API server.

./server.py                           # serves archive.db on :5273

Endpoints: /api/channels, /api/channels/:id/messages, /api/messages/:id, /api/search, /api/users. Also serves downloaded attachments from /attachments/.

Viewer (app/)

React + TypeScript + Vite application with:

• Virtual scrolling (TanStack Virtual) for large channels
• Full-text search with #channel and @user autocomplete
• Permalink URLs (#channelId/messageId)
• Message grouping, reactions with tooltips, embed rendering
• Keyboard navigation via use-kbd (Cmd+K omnibar, / search, ? shortcuts)
• Responsive layout (collapsible sidebar, mobile support)
• Prefetch on hover (channels, search results, mentions)

cd app
pnpm install
pnpm dev        # http://localhost:5272 (proxies /api to :5273)
pnpm build      # production build

Deployment

Cloudflare (Workers + D1 + Pages)

The api/ directory contains a Cloudflare Worker that serves the same API backed by D1.

cd api
pnpm install

# Create D1 database
npx wrangler d1 create my-discord-archive
# Update wrangler.toml with the database_id

# Import data
./d1-import.sh ../archive.db           # local D1
./d1-import.sh --remote ../archive.db  # remote D1

# Deploy worker
npx wrangler deploy

# Deploy viewer
cd ../app
VITE_API_BASE=https://your-worker.workers.dev pnpm build
npx wrangler pages deploy dist --project-name my-discord-archive

Incremental updates

./archive.py                    # fetch new messages
./build_db.py                   # rebuild SQLite
cd api && ./d1-sync.py --remote # sync delta to D1 (zero downtime)

GitHub Actions (for downstream repos)

This repo provides composite actions that downstream repos use in their workflows:

# Deploy the viewer to Cloudflare Pages
- uses: Open-Athena/discord-agent/actions/deploy-app@v1
  with:
    pages_project_name: my-discord-archive
    vite_api_base: https://my-worker.workers.dev
    cloudflare_token: ${{ secrets.CLOUDFLARE_TOKEN }}
    cloudflare_account_id: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}

# Deploy the Worker API
- uses: Open-Athena/discord-agent/actions/deploy-worker@v1
  with:
    cloudflare_token: ${{ secrets.CLOUDFLARE_TOKEN }}
    cloudflare_account_id: ${{ secrets.CLOUDFLARE_ACCOUNT_ID }}
    wrangler_toml: wrangler.toml  # your project's wrangler.toml

# Update the archive (fetch + rebuild + sync D1)
- uses: Open-Athena/discord-agent/actions/update-archive@v1
  with:
    discord_token: ${{ secrets.DISCORD_TOKEN }}
    cloudflare_token: ${{ secrets.CLOUDFLARE_TOKEN }}
    wrangler_toml: wrangler.toml

See actions/ for full input documentation.

DVX / Data versioning

The archive/ directory is tracked with DVX (a DVC fork). Each archive update creates a new snapshot; individual file blobs are deduplicated.

dvx add archive           # track archive state
dvx push                  # push to S3 remote
dvx pull                  # restore archive from remote

Discord bot setup

1. Go to the Discord Developer Portal
2. Create a new application, add a bot
3. Enable Message Content Intent under Bot settings
4. Generate a bot token → set as DISCORD_TOKEN
5. Invite the bot to your server with Read Message History + Read Messages permissions
6. Find your guild ID (right-click server name → Copy Server ID) → set as DISCORD_GUILD