Observations
Everything you set in the observation editor — the target, the data requests, and the observation-wide configuration — adds up to a single, self-contained description of what to observe and how. Skynet calls that description the observation spec.
You don’t have to think about the spec to use the editor. But knowing it’s there explains how a few things work, so it’s worth a short tour.
One description, built as you go
Section titled “One description, built as you go”As you move through the editor’s steps, you’re filling in one spec. The editor holds it for you and builds your observation from it when you finish. There’s just the spec you build and the observation you create from it.
A spec is deliberately portable: it describes the science (target, requests, settings) without naming an owner, a funding grant, or any internal IDs. The only references it carries are the public UIDs of any instruments you restricted it to. That’s what lets the same spec be reused in different places.
What that lets you do
Section titled “What that lets you do”- Duplicate an observation. Open any observation and choose Resubmit as new observation from the More actions menu — Skynet reads that observation’s spec and drops you into a fresh editor session pre-filled with the same setup. Tweak the target or settings and create a new observation, without rebuilding it from scratch.
- Export a spec. From an observation’s page, More actions → Export spec… shows its spec as portable JSON you can copy, keep, version, or hand to a colleague. Pasting it into Import from spec on the editor hub opens a fresh editor session from it (and keeps it as a draft).
- Save as draft. Not ready to commit? Save as draft saves your in-progress spec on Skynet so you can reopen it from the editor hub and finish later. It saves the spec, not an observation — nothing is scheduled, nothing is charged, and it won’t appear in your observation list. The editor also keeps your current work in your browser as you go, so a reload won’t lose it.
- Start from a ready-made spec. The project catalog and alert subscriptions both carry pre-built specs — the catalog’s pre-configured observations and an alert’s response observation are the same portable description, filled in for you.
Editing after you create the observation
Section titled “Editing after you create the observation”Because the editor works from the spec, editing an existing observation is the same flow as building a new one: the editor loads the observation’s current spec, you change what you need, and nothing reaches the live observation until you click Save changes on the Review step. See Review and publish → After you create it.
← The observation editor · 1. Observing accounts →
Observation model
Section titled “Observation model”flowchart LR O["Observation<br/>active · completed · canceled"] --> T["Target"] O --> CFG["Configuration<br/>one per observation type"] CFG --> BM["Brightness models<br/>target · saturation (optical)"] O --> DR["Data requests"] T --> POS["Position<br/>fixed · catalog object · orbital · ephemeral"] POS -. catalog object .-> CO["major body · planetary satellite<br/>MPC comet · MPC orbit · NORAD satellite · NEOCP"] DR --> REQ["Optical imaging · Radio tracking · Radio mapping"] DR --> RES["Results<br/>one per epoch × sample × copy"] RES --> TASK["Tasks<br/>one live task per result, on one telescope"]
The observation is built from a spec (target, configuration, data requests). Skynet expands the requests into results, and the dispatcher creates a task for each result on a specific telescope and instrument. See Results and tasks.
Observation lifecycle
Section titled “Observation lifecycle”sequenceDiagram autonumber participant U as User or alert participant API as Skynet API participant D as Dispatcher participant S as Telescope scheduler participant N as SkyNode participant P as Pipelines U->>API: Create observation (editor, API, or alert subscription) API->>API: Expand requests into results D->>API: Create a task for each open result on a telescope S->>API: Place each task in an observing window N->>API: Fetch the planned task when its window opens N->>N: Run the task on the instrument N->>API: Upload frames API->>P: Process (calibration, plate solve, photometry, stacks) P-->>API: Register results' files API-->>U: Notify and show files on the observation page Note over D,S: A task that misses its window slides to a later one, or is handed back so the dispatcher can create a new task