---

Afterfeed turns disconnected export files into one calm, searchable timeline. Your archive and media stay in the installation you control, original metadata is retained, and private annotations never modify the source records. Authentication and per-user ownership keep libraries isolated when an installation has more than one user.

✨ What you can do

• 🗂️ Browse everything together — normalize posts, replies, messages, and media from multiple services into a single timeline.
• 🔎 Find old moments — use full-text search, reconstructed threads, yearly heatmaps, statistics, maps, and a cross-service On This Day view.
• 🏷️ Curate privately — favorite, tag, annotate, collect, or hide posts without changing the imported archive.
• 🧭 Understand your history — explore activity patterns and a local relationship index built from conversations, follows, friends, mentions, and reactions.
• 🎁 Share selectively — turn one memory into a social-ready image, or export a collection as HTML, PDF, JSON, or a media bundle with granular privacy controls and metadata stripping.
• 🔌 Connect your tools — query your own archive through a user-scoped, read-only JSON API or the bundled Model Context Protocol server.

📦 Supported archives

Sensitive security and account data—such as IP history, device tokens, phone numbers, cryptographic keys, payment records, birth dates, and advertising preferences—is intentionally excluded where present.

🚀 Quick start

Requirements

• PHP 8.4.1 or later
• Composer
• Node.js and npm
• SQLite (default), MySQL/MariaDB, or PostgreSQL

Install

git clone https://github.com/rebeccathedev/afterfeed.git
cd afterfeed
composer run setup
composer run dev

Then open the local URL shown in your terminal and create the first account. SQLite works out of the box; see database backend setup to use MySQL/MariaDB or PostgreSQL with native full-text search. The first account adopts data from an older single-user database during an upgrade.

Install with Docker

Requirements: Docker Engine with the Compose plugin and access to the repository's GitHub Container Registry package. The application image contains PHP-FPM only; the included production Compose file adds Nginx and persistent volumes for SQLite, archived media, and compiled public assets.

git clone https://github.com/rebeccathedev/afterfeed.git
cd afterfeed

export AFTERFEED_IMAGE=ghcr.io/rebeccathedev/afterfeed:latest
docker pull "$AFTERFEED_IMAGE"
docker run --rm --entrypoint php "$AFTERFEED_IMAGE" artisan key:generate --show

Copy the generated key into a new .env.container file. Replace the image and URL values for your deployment:

AFTERFEED_IMAGE=ghcr.io/rebeccathedev/afterfeed:latest
APP_KEY=base64:paste-the-generated-key-here
APP_URL=http://localhost:8080
AFTERFEED_PORT=8080
AFTERFEED_MCP_ALLOWED_ORIGINS=http://localhost:8080

Start Afterfeed:

docker compose --env-file .env.container -f compose.production.yml up -d
docker compose --env-file .env.container -f compose.production.yml ps

Open http://localhost:8080 and create the first account. Keep .env.container private and backed up: changing APP_KEY invalidates sessions and encrypted application data. The entrypoint creates SQLite, applies migrations, refreshes public assets, and warms Laravel caches before FPM starts.

To update later:

docker compose --env-file .env.container -f compose.production.yml pull
docker compose --env-file .env.container -f compose.production.yml up -d

For HTTPS, external databases, backups, a custom FastCGI proxy, and troubleshooting, see container deployment.

Import an archive

Afterfeed automatically recognizes supported Twitter/X, Mastodon, Facebook, Reddit, Instagram, Google+, Nextdoor, LiveJournal, and Jekyll archive formats. Browser uploads are attached to the signed-in user. For a command-line import:

php artisan archive:import /path/to/twitter-export.zip me@rebeccapeck.org

Imports are safe to repeat. A content fingerprint prevents accidental duplicate imports, while re-importing an existing archive refreshes its normalized records through the latest importer so newly supported metadata can be backfilled.

The import guide documents format recognition, ownership, privacy, and the Google+ Takeout layout and coverage in detail.

🧭 Explore and curate

The timeline is only the beginning:

• Search uses SQLite FTS5 or the selected database's native full-text engine.
• Maps combine geotagged posts and privately annotated coordinates using locally bundled Leaflet; OpenStreetMap tiles load only when the map is opened.
• Statistics summarize activity by year, hour, weekday, service, and photo volume, plus common words, hashtags, Reddit communities, and frequently mentioned people.
• People creates a relationship directory from mentions, replies, direct messages, friends, follows, and attributable reactions.

Archive imports refresh the People index automatically. To rebuild it manually:

php artisan archive:people me@rebeccapeck.org

🎁 Export with control

Collections can become a self-contained HTML album, a PDF memory book, a JSON subset, or a ZIP containing HTML, JSON, and selected media. Choose whether to include text, media, identities, dates, private annotations, original links, and precise locations.

Privacy-friendly defaults remove image EXIF, remux video losslessly with FFmpeg to strip metadata, and omit raw platform metadata. Self-contained HTML and PDF media is capped at 60 MB, and PDF books at 250 posts; use ZIP for larger collections.

Individual posts can also become square, portrait, or landscape PNG cards. Identity, platform, date, footer, and archived photo controls make it easy to share an old memory without exposing or linking to Afterfeed.

🔌 API & MCP

Afterfeed includes a read-only JSON API under /api/v1:

GET /api/v1/status
GET /api/v1/accounts
GET /api/v1/posts?q=&account_id=&platform=&year=&month=&has_media=&page=&per_page=
GET /api/v1/posts/{id}
GET /api/v1/statistics

Results use bounded pagination of at most 100 records per page and omit hidden posts. Every request requires a per-user bearer token created in Settings → API & MCP access, and the API exposes only that user's archive.

To connect a local MCP client such as Claude Desktop or Codex, configure it to launch:

php /absolute/path/to/afterfeed/artisan afterfeed:mcp me@rebeccapeck.org

The stdio server provides search_posts, get_post, list_accounts, and archive_statistics. Remote clients can use the authenticated Streamable HTTP endpoint at /api/mcp; see remote MCP setup and security. Each token is revocable and can access only its owner's archive.

🧩 How it works

Every platform importer translates its export into four shared records:

SocialAccount  ──<  Archive  ──<  Post  ──<  Attachment
   identity         import        item         local media

• SocialAccount represents one identity on a platform.
• Archive represents one export and its source account.
• Post represents a normalized post, reply, message, or timeline item.
• Attachment represents local media associated with a post.

Original export metadata remains in JSON columns, keeping normalization lossless. Large Facebook exports use chunked database upserts and streamed ZIP media reads, avoiding full extraction and unbounded memory use.

🛠️ Development

# Start the application, queue worker, logs, and Vite
composer run dev

# Run the test suite
composer test

# Build production assets
npm run build

---