TelescopeYoke
A self-measuring robotic observatory for affordable GoTo mounts, run by a person and an AI agent together.
Free and open source. Runs on a laptop beside the telescope: Linux, with a Windows port.
What it does
You already own the mount, the handset and a camera. TelescopeYoke is the part a smart telescope has and yours does not. It is built around one idea: measure the imperfections of modest equipment instead of pretending they are not there.
- Home position set by eye? It photographs the sky, plate-solves where the telescope really ended up, and corrects until the target is in the middle. On its first real night an EQ3 that started about ten degrees out centred M27.
- Manual focuser? Keep your hand on it and listen: each frame's star size is said as a number, as used under real stars. With
--tones, a click when your last turn has been measured, a tone that rises as the stars sharpen, and "Focus good. Hold" at the end; those have run on the demo's stars only so far. - House and trees in the way? Give it a phone panorama and it learns your garden's skyline, so the planner stops recommending targets you cannot see.
- Polar alignment rough? Three plate solves measure it and say which way to turn each adjuster , and an experimental mode tries it before dark from Polaris alone.
- Takes and stacks the pictures: short exposures, each one checked, poor ones rejected, the rest lined up and added, with every raw frame kept so the picture can be rebuilt later.
- Plans the night: darkness, Moon, cloud, and a GO / MARGINAL / NO-GO verdict, with targets ranked for your own sky.
- Controls Sky-Watcher SynScan mounts through the handset's serial lead, correcting for a handset clock that is wrong.
- Keeps the mount inside its limits. Every move is checked against the limits first. A move you make in the application is shown as a plan and goes only when you confirm it; a night's schedule you have agreed runs its own moves through the same checks.
- Runs the whole night: the agreed steps in order, a failed step fetches you, and at the end it parks the mount and tells you. The steps and their GoTos have run on the real mount; the night's end (parking and telling you) has not yet run by itself through a real night.
- Hunts for something new: novae in M31 and M33 by image subtraction, exoplanet transits; each find checked before it reaches you, with the evidence.
- Researches without the telescope: Kepler and TESS light curves searched for stars that dim like Tabby's, each lead checked against Gaia, SIMBAD and VSX. No discovery yet, and it says so.
- Runs a cooled mono camera and filter wheel: cooled slowly once and held cold all night, several filters and targets in one cold session. Its first colour picture failed; the fix is not yet rerun.
- Tells you what matters: on your phone with the evidence, and when it needs your hands, through the house's lights and a speaker.
- Explains itself live: a YouTube and Twitch broadcast with the stack building up, a presenter that says what and why, and chat commands viewers can ask but not operate.
- Works with an AI agent: every command answers in a fixed JSON shape, skills and specialist agents hold the routines, and code that does not depend on the agent decides what may move.
It runs on your own computer. No account, no cloud service.
Real results
Already using NINA and PHD2? Analyse last night without changing a thing
The companion reads the frames your capture program saved and PHD2's guide log, measures every frame against that night's best of the same setup, and says where the time went. It touches no device. What it measured is kept apart from what it only infers, and every file is accounted for. Here is a real night (the author's own, 5 November 2026), trimmed:
706 files: 693 light frames measured; 13 left out, no single-plane image (a stack or a colour picture).
M33 L (183m, gain 1500, 16 s) 00:23-01:41: 57 min measured; 37 min acceptable by its own best frames (139 of 213 frames)
15 min (56 frames) dimmed: likely cloud or haze
5 min (18 frames) trailed: likely the mount's tracking or the wind (no guiding to compare)
No frames 01:15-01:33 (18 min beyond the usual gap): not known: nothing it reads says
Where the time went: no frames (explained) 44 min, dimmed 38 min, no frames (unexplained) 36 min, trailed 11 min, soft 6 min
In the download since 0.4.1. Tried on the author's frames, on made-up nights with known faults, and offline on NINA and PHD2 logs other observatories have published (it told one night's 27 ten-minute lights of M 27 from its autofocus, plate-solving and focusing frames, judged them by NINA's own star detection, and found a five-minute stretch where PHD2 said it was guiding with no guide step); not yet live beside anyone else's observatory. More, with another observatory's night and how to run it.
Try the demo: no telescope needed
The demo is the whole application on a pretend telescope. Nothing is connected to your mount and nothing can move, so it is safe to try on the computer that sits beside real equipment.
Ubuntu or Debian
wget https://github.com/Ryan-Clinton/telescopeyoke/releases/download/v0.4.2/TelescopeYoke-v0.4.2.zip
unzip TelescopeYoke-v0.4.2.zip
cd TelescopeYoke-v0.4.2
./install.sh --demo
Windows 10 or 11
- Install 64-bit Python 3.11 or newer from python.org, ticking "Add python.exe to PATH".
- Download TelescopeYoke-v0.4.2.zip and unpack it.
- Double-click
try-demo.cmd(not yet run on a real Windows machine: the Windows camera has taken frames, the installer and window have not been tried).
A window opens with tonight's report. Pick a target, press "Centre with plate solve", and watch it find and correct its own pointing error. Then try Focus and Imaging; "The pretend sky" lets you bring cloud over or pull the camera's lead out, to see what it does when a night goes wrong.


Supported hardware
It is new, and it says plainly what it has and has not been proven on. So far it has run on one telescope, the author's:
| Mount | Handset | Camera | System | What has worked | Tried by |
|---|---|---|---|---|---|
| Sky-Watcher EQ3 Pro | SynScan, firmware 3.35 | Altair Hypercam 183C | Ubuntu 26.04 | Planner, mount moves, camera, focusing, plate solving and GoTo with centring, under real stars | the author |
| the same EQ3 Pro | the same | the same | Windows 11 | The demo, and camera frames indoors. No mount driven yet | the author |
| the same EQ3 Pro | the same | Altair Hypercam 183M PRO TEC (mono, cooled), Starlight Xpress filter wheel, 0.9x coma corrector | Ubuntu 26.04 | First night 4-5 November 2026: cooled to -5 °C, darks, every filter, plate solving, polar alignment, M33. No flats yet | the author |
| EQ5 | SynScan | any | Linux or Windows | wanted | could be you |
| HEQ5 | SynScan | any | Linux or Windows | wanted | could be you |
| EQ6, EQ6-R | SynScan | any | Linux or Windows | wanted | could be you |
| a mount with an EQStar or other EQMOD-style controller | none | any | Linux or Windows | wanted: whether it answers at all | could be you |
EQ5, HEQ5 and EQ6 mounts with a SynScan handset speak the same serial protocol, so they are likely to work, but nobody has tried. On Linux the camera is read through INDI; on Windows, Altair cameras through their own library. New in 0.4.1: a mount, camera, focuser and filter wheel through their ASCOM drivers (Alpaca), tried only on ASCOM's own simulator so far; guiding by PHD2, tried only against a pretend PHD2; and a companion that watches an observatory you already run, from the frames it saves and PHD2, without touching it, run on the owner's own 693-frame night and on published NINA and PHD2 logs of three other observatories, not yet live beside anyone else's.
./doctor.py --report only looks at what is connected: it writes out what the handset says the mount is, its firmware, and what was found, with your location left out. Paste that into a
hardware report, or ask first in
Discussions. Your name goes beside your row if you want it there.
Find out more
- From box to first galaxy: a first night, step by step, for someone new to motorised telescopes. No knowledge of polar alignment, plate solving or astrophotography assumed.
- The README: everything it does.
- By what you came for: NINA and PHD2 night analysis, SynScan automation, ASCOM Alpaca.
- Results so far: each real result with its method and its limits, from the first self-centring GoTo to the cooled camera's first night.
- All the documentation, sorted by what you came for.
- What has been proven: what works on real hardware, what is partly proven, and what has run only on the simulator.
- Setup: installing it for a real telescope, on Ubuntu and on Windows.
- Changes in each release.
- Discussions: questions, ideas, and pictures you have taken.