Verbinal for macOS
A native macOS desktop companion for the CANFAR Science Portal, built with SwiftUI and XcodeGen
— and, since 1.4.1, on iPhone and iPad too.
Project home: verbinal.com
Available on the Mac App Store,
and as an unsigned Mac build from GitHub Releases.
For iPhone and iPad, it is on the App Store.
User manual: read it online, or here in the repository:
English · Français. It covers every
screen, setting and shortcut of Verbinal 1.4.
Features
-
On iPhone and iPad (new in 1.4.1) — check, review and verify away from the Mac. The
FITS and Cube viewers and the files on the device need no account: open a file from Files,
Mail, AirDrop or your storage, header first, check its CHECKSUM, WCS and spectral axis, and
share it. With a CANFAR account, the Portal and Compute: your sessions, batch jobs and runs.
It is not a port of the Mac app; everything below is the Mac's.
-
Sessions — launch and manage Notebook, Desktop, CARTA, Contributed, and
Firefly sessions on the CANFAR Skaha platform, no browser required. Auto-refreshing
status, live platform load, CPU and RAM (and GPU, when there is one) on every session
card, a GPU for a session or a batch job when CANFAR offers them, one-click re-launch
from history, in-place event and container logs, and headless batch jobs that keep up
with thousands of jobs — with a history kept after the platform drops them.
-
Remote Compute (new in 1.4) — run Python or Bash on a session on your own CANFAR
account, from the app or through your AI assistant. Every run and its output are kept,
and a run survives signing out and quitting the app.
-
CADC archive search — build queries against the CADC TAP service (with a radius for
cone searches and VizieR beside CADC), review results in a sortable, filterable table,
switch units (RA/Dec HMS·DMS, 14-unit spectral conversion), and open rich CAOM2
observation detail — with an ADQL editor that checks your query as you type. Every row
of a query of products has its publisher ID, so previews, details and downloads work for
your own ADQL too.
-
Research assistant — a workspace that tracks your observations, with or without
their files: download whole observations, choose individual files, or cut out just the
region you need. Keep each file on this Mac, in your CANFAR storage, or both, and move
it with Move To…. Full-text search over notes and tags, and bundle export.
-
VOSpace storage — a native file browser to browse, upload, organise, share and
manage your VOSpace files, with quota and usage at a glance and a warning when a
sensitive file is public.
-
Your work on every device (new in 1.4.1) — choose, part by part, what Verbinal keeps
in your CANFAR storage: Research, marks on images, workflows, saved queries, guides for
assistants and more, the same on every Mac you sign in on. Changes are kept by
revision, never by a clock, and a conflict waits for you to choose.
-
FITS viewer — hardware-accelerated, Metal-based rendering with pan/zoom, scaling
modes, WCS-aware pixel readout, and full zenithal projections (TAN/SIN/STG/ZEA).
Marks on your images kept with the file and exported to DS9, plotted spectra (such as
HST x1d), and figures of the whole image or a region.
Screenshots
The 3D Cube Viewer (new in 1.3) — explore FITS spectral cubes as an interactive, GPU ray-marched volume:
| Home (1.4) | CADC archive search |
|---|
| |
| Portal (1.4.1) — CPU, RAM and GPU load | Research (1.4.1) — each file on this Mac and in your storage |
|---|
| |
| FITS viewer | AI assistant (MCP) driving Verbinal |
|---|
| |
Installation
Download
Download the latest .dmg from GitHub Releases.
- Open
Verbinal-macOS.dmg
- Drag Verbinal to Applications
- On first launch, right-click the app and select Open (macOS Gatekeeper requires this for unsigned apps)
A .zip archive is also available if you prefer.
Verify your download with the checksums-sha256.txt file:
shasum -a 256 -c checksums-sha256.txt
iPhone and iPad
Install Verbinal from the App Store.
Build from source
See Building below.
Requirements
Runtime
- macOS 14.4 or newer, or iOS and iPadOS 17 or newer
- A CANFAR account (on iPhone and iPad, only for the Portal, Compute and your storage)
Build
- Xcode 16 or newer
- XcodeGen 2.45 or newer
Building
# Generate the Xcode project
xcodegen generate
# Debug build
xcodebuild build \
-project Verbinal.xcodeproj \
-scheme Verbinal \
-destination 'platform=macOS' \
-derivedDataPath .derivedData \
CODE_SIGNING_ALLOWED=NO
# Run tests
xcodebuild test \
-project Verbinal.xcodeproj \
-scheme Verbinal \
-destination 'platform=macOS' \
-derivedDataPath .derivedData \
CODE_SIGNING_ALLOWED=NO
For local development, you can also open the generated Verbinal.xcodeproj in Xcode and run the Verbinal scheme directly.
Running Tests
xcodebuild test \
-project Verbinal.xcodeproj \
-scheme Verbinal \
-destination 'platform=macOS' \
-derivedDataPath .derivedData \
CODE_SIGNING_ALLOWED=NO
Code Quality
- All source files include MPL-2.0 license headers
- CI runs build and test on every push and pull request
- Unit tests cover URL construction, model mapping, networking, XML parsing, and image parsing
- Strict separation of concerns: views, view models, services, and models
- A single external dependency, pinned to an exact version: GRDB 7.11.0 (in-app SQLite store); everything else is Apple frameworks
Project Structure
project.yml # XcodeGen source of truth
Verbinal/ # Application code, models, services, views
VerbinalTests/ # Unit tests for parsing and other pure logic
Architecture
- SwiftUI for the UI layer
- Observation-based app state and view models
- Async/await networking with
URLSession
- Clear separation between models, services, view models, and views
API Endpoints
All communication is with CANFAR services over HTTPS. No telemetry, analytics, or third-party calls are present.
| Service | Base URL | Purpose |
|---|
| Auth | ws-cadc.canfar.net/ac | Login, token validation |
| User info | ws-uv.canfar.net/ac | User profile retrieval |
| Sessions | ws-uv.canfar.net/skaha/v1 | Session CRUD, images, context, stats |
| Storage | ws-uv.canfar.net/arc | VOSpace quota |
License
Mozilla Public License 2.0
Copyright (C) 2025-2026 Serhii Zautkin
Privacy
See PRIVACY.md. In short: no data collection, no telemetry, and no third-party services. Data stays on your machine or goes directly to CANFAR.
Related projects
These open-source companions help you get the most out of Verbinal:
- verbinal-execution — a
CANFAR/Skaha contributed-session image that powers Verbinal's AI remote compute
(
run_code). It is a file-drop watcher that runs agent-supplied Python/bash snippets
and writes JSON results back — no shell, no inbound network. Run it in a contributed
session to let your AI assistant execute code on the platform on your behalf.
- inspector-image — a minimal
Alpine container image with Anchore syft preinstalled,
used as the Skaha/CANFAR inspector that powers Verbinal's image content discovery
(probing container images for their Python, R, system, and OS-level packages).
Contributing
See CONTRIBUTING.md.
