Filefy — A Self-Hosted, Web-Based Modern Cloud File Manager

pip install filefy && filefy

Open your browser, manage files from anywhere, download from remote URLs, compress/extract archives, transfer files to peer servers, and share a public URL via Cloudflare Tunnel — all without leaving the browser and without installing anything extra on the client side.

The Problem

Every developer who manages a remote server, a NAS box, or a cloud VM eventually reaches for the same tired toolkit: scp, rsync, wget, raw tar commands, and occasionally a heavy FTP client. The workflow is friction-heavy, requires CLI knowledge on both ends, and leaves non-technical collaborators completely stranded.

Filefy solves this by shipping a complete, production-ready file manager as a single Python package that runs in any browser.

What Is Filefy?

Filefy (v1.1.1, MIT License) is a professional, self-hosted web file manager written in Python. It runs a Flask server and serves a polished dark-theme single-page application. Every operation — uploading, downloading, compressing, extracting, remote-fetching — happens server-side, while the browser acts purely as a control interface.

Key design principles:

• Zero client-side dependencies — users need only a modern browser.
• Zero mandatory cloud accounts — everything runs on your own machine.
• Background-task architecture — every long-running job runs in a daemon thread; the browser polls a progress endpoint and shows a real progress bar.
• Minimal Python footprint — the only runtime dependencies are Flask, Werkzeug, and Requests.

Highlighted Features

1 · Resumable Chunked Upload

Uploads use a three-step protocol (upload-init → upload-chunk → upload-complete) that streams files in configurable chunks. If the connection drops mid-way, the client resumes from the last acknowledged byte — no duplicate work.

POST /api/upload-init      → create session, receive upload_id
PUT  /api/upload-chunk/<id> (Content-Range: bytes start-end/total)
POST /api/upload-complete/<id>
GET  /api/upload-status/<id>   → bytes received so far (resume point)
DELETE /api/upload-cancel/<id>

Controls in the Transfer Center: Pause · Resume · Cancel.

2 · Remote URL Download with Progress

Paste a URL, click New Download. Filefy fetches the file on the server side (using requests with streaming), stores it in the current directory, and streams back byte-count progress. Multiple downloads run concurrently.

POST /api/remote-download    → { "urls": ["https://…"] }  → task_ids[]
GET  /api/download-progress/<task_id>
POST /api/pause-download/<task_id>
POST /api/resume-download/<task_id>
POST /api/cancel-download/<task_id>
POST /api/dismiss-download/<task_id>

All status updates are displayed in the Transfer Center, which can be minimised to the sidebar.

3 · Compress Archives (zip / tar / tar.gz)

Right-click any file or folder → Compress as… → choose format and name. A daemon thread builds the archive while the browser polls for progress.

POST /api/compress          → { "sources": […], "format": "zip|tar|tar.gz" }
GET  /api/compress-progress/<task_id>
POST /api/cancel-compress/<task_id>

Supported formats: .zip, .tar, .tar.gz (with optional pigz for faster gzip). Large multi-GB trees are handled without blocking the server.

4 · Extract Archives — "Extract Here" (NEW)

The mirror image of compression. Right-click a recognised archive file → Extract Here. The same background-task + progress-polling pattern applies, so the Transfer Center shows a live progress bar with extracted-bytes / total-bytes and current filename.

Supported input formats: .zip, .tar, .tar.gz, .tgz, .tar.bz2, .tar.xz, .gz, .bz2

POST /api/extract           → { "path": "/abs/path/archive.zip" }
GET  /api/extract-progress/<task_id>
POST /api/cancel-extract/<task_id>

The Extract Here context-menu item appears only when the selected file has a recognised archive extension — it stays hidden for all other file types.

5 · Cloudflare Tunnel — Public URL on Startup

Filefy automatically starts a Cloudflare Quick Tunnel when it launches. Within seconds, both the local URL and a free *.trycloudflare.com public URL are printed to the terminal:

═══════════════════════════════════════════════════════
   FileFy v1.1.1 - Web-Based File Manager
═══════════════════════════════════════════════════════
   Base Directory: /home/user
   Local URL:      http://0.0.0.0:5000
   Public URL:     https://random-name.trycloudflare.com
═══════════════════════════════════════════════════════

The public URL is also surfaced in the UI so you can share it instantly. If you do not need public access, pass --no-tunnel to skip it.

filefy              # local + public URL (default)
filefy --no-tunnel  # local only

cloudflared is installed automatically on first use if it is not already on PATH. No Cloudflare account required.

6 · Server Bridge — Peer-to-Peer File Transfer Between Servers

This is arguably Filefy's most unique capability. The Server Bridge lets two independent Filefy instances connect to each other and exchange files — entirely through the browser, with no scp, no VPN, and no shared storage.

The Problem It Solves

Copying files from Server A → Server B traditionally means:

1. SSH into Server A, scp/rsync to Server B, or
2. Download to your laptop, then re-upload to Server B.

With the Server Bridge, you open Filefy in a single browser tab and orchestrate the transfer directly between the two servers — your laptop never touches the data.

Step-by-Step: Connecting Two Servers

On Server A (the server being connected to):

GET /api/bridge/generate-code?url=https://server-a.example.com

This returns a compact, URL-safe Base64 pairing code that encodes { "url": "...", "token": "<uuid>" }. The token is one-time use and expires after a configurable TTL (default: a few minutes).

{ "code": "eyJ1cmwiOiAiaHR0cHM6Ly9zZXJ2ZXItYS5leGFtcGxlLmNvbSIsICJ0b2tlbiI6ICIuLi4ifQ==" }

On Server B (the initiating side), paste the code:

POST /api/bridge/connect
{ "code": "<paste code here>", "name": "My Server B" }

Server B decodes the pairing code, calls Server A's handshake endpoint:

POST https://server-a.example.com/api/bridge/handshake
{ "token": "<uuid>", "peer_name": "My Server B" }

Server A:

1. Validates and consumes the token (one-time use — replays are rejected with HTTP 403).
2. Issues a session token (UUID) scoped to Server B.
3. Returns the session token plus Server A's hostname.

From this point on, Server B authenticates every peer-facing request with:

Authorization: Bearer <session_token>

Both sides now show each other in their Connected Peers panel.

Full Bridge API Surface

Push: Sending Files to a Peer

POST /api/bridge/push
{
  "peer_id":     "<uuid>",
  "files":       ["/abs/local/path/file1.tar.gz"],
  "destination": "/remote/dest/dir"
}
→ { "task_id": "<uuid>", "message": "Push started for 1 file(s)" }

The push runner:

1. Initiates a chunked-upload session on the peer (receive-init).
2. Streams the file in fixed-size chunks using Content-Range headers — the same protocol as a normal browser upload.
3. Finalises the session (receive-complete).
4. Repeats for every file in the list.
5. Reports bytes transferred, current filename, and speed in real time.

The transfer can be cancelled at any point; the partial file on the remote is cleaned up automatically.

Pull: Fetching Files from a Peer

POST /api/bridge/pull
{
  "peer_id":     "<uuid>",
  "files":       ["/remote/path/report.zip"],
  "destination": "/local/dest/dir"
}
→ { "task_id": "<uuid>", "message": "Pull started for 1 file(s)" }

The pull runner streams the remote file using requests with stream=True, writing it to the local filesystem in chunks. Progress (bytes, speed, current filename) is updated every 500 ms.

Remote File-System Operations

Beyond transferring files, the bridge can proxy file-system operations to a peer so you can manage the remote server as if it were local:

POST /api/bridge/remote-op
{
  "peer_id": "<uuid>",
  "op":      "delete" | "rename" | "create-folder" | "copy" | "move",
  "payload": { ... }
}

The payload is forwarded verbatim to the peer's regular API endpoint (/api/delete, /api/rename, etc.) and the response is returned as-is.

Security Model

• The pairing code is a short-lived, one-time token — intercepting an already-used code grants no access.
• All peer-facing endpoints require Authorization: Bearer <session_token>. Missing or invalid tokens return HTTP 401.
• File paths sent by peers are validated through the same get_safe_path() guard that protects the normal file manager — path-traversal attacks are blocked at HTTP 403.
• Sessions persist only in memory; restarting the server invalidates all active sessions.

Typical Use Cases

7 · Settings Modal

Click the ⚙ icon in the header to open the Settings modal. Current values are loaded from the server (GET /api/settings) and saved back (POST /api/settings) with instant feedback. Configurable fields:

8 · Rich File Operations

Installation

PyPI (Recommended)

pip install filefy
filefy

pipx (Isolated Environment)

pipx install filefy
filefy

From Source

git clone https://github.com/Pymmdrza/filefy.git
cd filefy
pip install -e .
filefy

Docker (One-Liner)

docker run -d \
  --name filefy \
  -p 5000:5000 \
  -v "$PWD/data:/data" \
  ghcr.io/pymmdrza/filefy:latest

Pre-built multi-arch (linux/amd64, linux/arm64) images are published on every release to ghcr.io/pymmdrza/filefy. The Docker image includes cloudflared pre-installed so the tunnel works out-of-the-box.

Docker Compose

services:
  filefy:
    image: ghcr.io/pymmdrza/filefy:latest
    restart: unless-stopped
    ports:
      - "5000:5000"
    environment:
      FILEFY_HOST: "0.0.0.0"
      FILEFY_PORT: "5000"
      FILEFY_DIR: "/data"
    volumes:
      - ./data:/data

docker compose up -d

CLI Reference

usage: filefy [-h] [-H HOST] [-p PORT] [-d DIR] [--debug] [--no-tunnel]
              [--install-cloudflared] [-v]

options:
  -H, --host HOST         Network interface to bind  (default: 0.0.0.0)
  -p, --port PORT         TCP port                   (default: 5000)
  -d, --dir DIR           Root directory             (default: home directory)
  --debug                 Enable Flask debug mode
  --no-tunnel             Skip the automatic Cloudflare public URL
  --install-cloudflared   Download & install cloudflared binary, then exit
  -v, --version           Print version and exit

Common patterns:

filefy                            # LAN + public Cloudflare URL
filefy --no-tunnel                # LAN only
filefy -p 8080 -d /srv/data       # custom port and directory
filefy --host 127.0.0.1           # localhost-only (most secure)
filefy --install-cloudflared      # one-shot cloudflared setup

Python API

Filefy exposes a programmatic interface for embedding in larger applications:

from filefy.server import run

# Minimal — binds to 0.0.0.0:5000, serves home directory
run()

# Custom configuration
run(
    host="127.0.0.1",
    port=8080,
    base_dir="/var/shared",
    tunnel=False,          # disable Cloudflare tunnel
    debug=False,
)

WSGI Integration

from filefy import create_app

# With Gunicorn
# gunicorn "filefy:create_app(base_dir='/data')" -b 0.0.0.0:8000 -w 4
app = create_app(base_dir="/data")

# Gunicorn
pip install gunicorn
gunicorn "filefy:create_app()" -b 0.0.0.0:8000 -w 4

# Waitress (cross-platform, great on Windows)
pip install waitress
waitress-serve --listen=0.0.0.0:8000 filefy:app

Architecture at a Glance

Browser (SPA)
    │  REST + JSON
    ▼
Flask Server (filefy/server.py)
    ├─ /api/browse            Directory listing
    ├─ /api/upload-*          Chunked resumable upload protocol
    ├─ /api/download          Local file download (Range support)
    ├─ /api/remote-download   Background remote fetch + progress
    ├─ /api/compress          Background archiving + progress
    ├─ /api/extract           Background extraction + progress
    ├─ /api/bridge/*          Peer-to-peer file transfer
    ├─ /api/settings          Runtime configuration GET/POST
    └─ /api/server-info       Version + tunnel status
    │
    ├─ Background threads (daemon)
    │   ├─ compression_task_runner
    │   ├─ extraction_task_runner
    │   ├─ remote_download_task
    │   ├─ bridge_push_runner     (one per push task)
    │   └─ bridge_pull_runner     (one per pull task)
    │
    └─ CloudflareTunnel (subprocess: cloudflared)

Every long-running operation follows the same pattern:

1. POST starts the job → returns task_id immediately (HTTP 200).
2. GET progress endpoint polls at ~750 ms intervals → returns { status, processed, total, … }.
3. POST cancel endpoint cooperatively cancels the running thread.
4. The Transfer Center in the UI renders all tasks in a unified panel.

System Requirements

Security Notes

• Filefy is designed for trusted networks (LAN, VPN, local machine). It does not ship with authentication by default.
• The --host 127.0.0.1 flag restricts access to the local machine only — recommended when running on a shared host without a reverse proxy.
• All file-operation endpoints validate that resolved paths stay within the configured BASE_DIR; path-traversal attempts are rejected with HTTP 403.
• The Cloudflare Tunnel URL is ephemeral — it changes every time you restart filefy and cannot be predicted.

Contributing

1. Fork github.com/Pymmdrza/filefy
2. Create a feature branch: git checkout -b feature/my-feature
3. Install dev dependencies: pip install -e ".[dev]"
4. Run the test suite: pytest
5. Open a Pull Request

Links

Built by Mmdrza · MIT License · Python 3.9+