Passes: what is coming and what happened
Passes holds every pass, ahead or behind. Upcoming answers what the window ahead will deliver: the GB your passes should carry, the passes that could fall short, and the recommendations that wait on a decision. Past answers how they did: the GB they delivered against the GB they could have, how often a pass landed inside its forecast range, and what was decided.
The page is three panes side by side. Each scrolls inside itself; the page does not.
| Pane | Upcoming | Past |
|---|---|---|
| Passes, on the left | Upcoming or Past, then the filters At risk, Waiting, All and the passes in the window | Upcoming or Past, then the filters All, Off range, Missed and the passes in the window, with Export CSV at its foot |
| The view, in the middle | Timeline: three cards, then the passes on a time axis | Live or Audit, three cards, and the open card's chart |
| On the right | The pass you chose, or Waiting on you | The pass you chose, or Recently decided |
Upcoming and Past live in the address (?when=past; Upcoming is the default), so a link opens the same side. Switching sides starts the other one at its defaults.
The context and Share
The view's title row holds the selects that set what the side reads, and Share. They live in the address, so a link opens the same view.
| Control | Upcoming | Past |
|---|---|---|
| Workflow | The workflows that price passes, shown when you have more than one; default the pass-yield workflow | The audit workflows, shown when you have more than one |
| Scope | Whole fleet, one ground station, one antenna or one satellite | The same |
| Window | Next 12 hours, Next 24 hours, Next 48 hours (the default), Next 72 hours | Last 24 hours, Last 48 hours, Last 7 days (the default). A review reaches back at most 7 days: one run prices at most 10,080 steps, 7 days at the one-minute step |
A scope narrows every figure, the chart and the list. The cross beside the selects clears it. Share copies the view's link or its API call; see Share a view.
Upcoming
Everything on this side comes from the newest run of your pass-yield workflow, with your other workflows laid over it.
Price the window
Upcoming reads runs; it does not start them on its own. When the newest run has not finished, failed, or does not cover the window, a notice at the head of the Timeline pane says so and offers Price the next 48 hours (the label follows the window). That starts a test run of the workflow over the window: it prices every pass and delivers nothing, and the button shows the run's step (Finding passes, Asking the SNR model, and so on) until it settles.
Test runs are open on every plan. On Explore a run covers at most 2 satellites of your own; the demo fleet's never count. When the fleet holds more of your own, the run is refused and Choose 2 satellites opens, with the 2 that have the most passes in the window already ticked. Change the pick if you like; Save and price saves it on the workflow's input as its next version and prices the window. Nothing is saved until you choose it. A line at the foot of the Timeline pane then says Priced for those satellites, and that Explore prices up to 2.
The passes ahead
The list's title row counts the passes that are fine and those at risk ("38 fine · 4 at risk"). Under it, three filters on one line, At risk and Waiting with their counts (the title row already totals All); hover a filter's words for what it holds:
- At risk: a pass that could fail, that a step could not price, that is under your contract minimum in its low case, or whose recommendation still waits on you. A recommendation already decided is under Past, not here.
- Waiting: passes whose recommendation waits on your decision, nearest deadline first.
- All: every pass in the window.
Each row is two lines. The first has a dot (amber needs a look, green is fine, gray is not priced), the satellite and the antenna by name, and the expected GB at the right; hover it for the likely range. The second has the start time and, where there is one, the recommendation in a few words with its deadline ("Swap to Troll, in 2 h"), or what puts the pass at risk. Past 20 rows a search box narrows by satellite, ground station or pass. Choose a row to open the pass.
The Timeline pane
Three cards lead it:
| Card | Shows |
|---|---|
| Expected | The expected GB the fleet can carry over the window, and under it the floor: the total if every pass had its low case at once. A satellite two antennas hear at the same time counts once, an antenna's overlapping passes count its best schedule, and a pass running past the window (a day slot of a GEO link) counts only its part inside it |
| At risk | Passes that could fail (a low case of 0 GB), that a step could not price, or that fall under your contract minimum in the low case, out of all passes in the window |
| Waiting on you | Recommendations held for a decision across every workflow, and how soon the nearest one is due. The card chooses the list's Waiting filter |
Set a contract minimum on the At risk card opens a dialog that sets the GB each pass must deliver in its low case. It opens filled with the median pass, so Save minimum is one click; Change and Clear adjust it later. The minimum is stored for the workspace, so everyone in it sees the same one (GET and PUT /predictions/contract-minimum).
Under the cards, the timeline draws each priced pass in the window on a time axis. By ground station and By satellite draw one row per ground station or per satellite; past a few rows the rows scroll inside the timeline. Choose a pass to open it. The line at the foot of the pane names the workflow and version, the network config version, the models that priced it and when, and says when the run was a test run.
Waiting on you
With no pass chosen, the right pane lists every recommendation held for a decision, across every workflow, nearest deadline first: how soon it is due, what it changes, and the pass. It refreshes while the page is open and lists the nearest 50. Choose one to open its pass.
A pass
A chosen pass fills the right pane, linkable as ?pass=<id>. Escape, or the close in its title row, goes back to Waiting on you.
- its name, then one sentence that answers for it from its figures and risks: "Likely 1.9 GB; could fail if the rain over Awarua holds.", "Likely 2.4 GB; its low case, 1.2 GB, is under the 2.0 GB minimum." or "Likely 6.8 GB; even its low case, 5.1 GB, clears the 2.0 GB minimum.";
- when it flies, and how soon;
- Expected: its likely range drawn against your contract minimum, with the expected GB under it;
- Recommendation, when it has one: what it changes, the backup pass, the gain in the low case and, while it waits, the deadline, with Send to MCS, which sends it to your mission control system (MCS), and Decline; see Deciding one;
- Why: the breakdown, each step by name (Link SNR, Weather, Capacity, Expected GB), its value and where the value came from (Physics, Weather, Calculated). Hover a step or a source for its name in the workflow. Reading a result explains it.
Its foot has Copy link, Copy API call (GET /predictions?workflow=<slug>&pass_id=<id>) and, when a COSMOS bridge has checked in, Open in COSMOS.
Before there is anything to show
| Upcoming says | Do this |
|---|---|
| No fleet on this account yet | Load the demo fleet, or Start from the catalog; see Demo fleet. With the demo fleet loaded, Upload 7 days of pass logs goes to the audit under Past |
| No workflow prices your passes yet | Start from a template opens the template picker; see New automation |
| The workflow has not finished a run | Price the next 48 hours |
Past
Everything on this side comes from the audit-ledger workflow, which sets each priced pass against what was logged for it.
Live or Audit
The switch at the head of the report picks the data:
| Reads | Use it for | |
|---|---|---|
| Live | Your live telemetry, through runs of the audit-ledger workflow | How your fleet is doing |
| Audit | A pass log you upload | A fleet that does not send live telemetry yet, or a past week to check; see Free audit |
Switching drops the run a link pinned and the breakdown it showed.
Run a review
With Live, Past shows the newest finished live run of the audit workflow. When there is none, or the newest one read other data, failed or was stopped, it says so and offers Review the last 7 days (the label follows the window). With no audit workflow yet, the same button saves the audit-ledger template as yours, with only its API output, and runs it. When the newest review covers another window, a notice says which, with the same button.
A review is a run that delivers, so it needs Operator or above. On Explore, Live says so in one sentence, with Open Audit, which Explore can run, and Compare plans. With no fleet yet, Past shows the same No fleet on this account yet as every other tab.
The passes behind
The list's title row counts the passes delivered and those not ("201 delivered · 30 not"). Under it, three filters on one line, Off range and Missed with their counts (the title row already totals All): All, Off range (the passes that delivered below or above their likely range, the furthest out first) and Missed (failed, missed, also a pass whose telemetry never showed lock, or with nothing logged). Hover a filter's words for what it holds.
Each row is two lines. The first has a dot (green delivered, red failed or missed, gray nothing logged), the satellite and the antenna by name, and the GB delivered at the right; hover it for the forecast. The second has when the pass flew and its expected GB, or how far outside its range it landed ("1.0 GB below its range"). Past 20 rows a search box narrows the list. Export CSV, at the list's foot, saves every pass with its forecast, what it delivered and its status. Choose a row to open the pass.
The three cards
Each card is also the switch for the chart below it, and names the report's pane.
| Card | Shows | Its chart |
|---|---|---|
| Delivered | The share of possible GB the passes delivered, and both totals | Delivered against possible, grouped by Ground station, Satellite or Day, under one line counting the passes delivered and dead. Possible GB counts each antenna's best schedule once where its passes overlap |
| In range | The share of passes with a forecast and a logged result that delivered inside their likely range, against the 80% aim | In range by Model or Ground station, each a bar against the 80% aim, under a line saying how many landed below and above their range (it opens Off range in the list) and the typical miss in GB |
| Recommendations (Live only) | The low-case gain the recommendations you followed forecast, added up | Each decided or recorded recommendation: the pass, what it changed, the decision (Followed, Declined, Expired, Trial, not sent, Not sent), who decided it and when, and its low-case gain |
The API does not report what a followed recommendation actually delivered, so the third card adds up the forecast gain, and says so in its tip. The line at the foot of the report names the workflow and version, the network config version, the models and when it was priced.
Audit
With Audit and no report yet, the report's pane holds two steps: Upload a pass log (or Use a sample pass log on the demo fleet), then Run the audit, filled with the last 7 days, or with the days the uploaded pass log covers once you upload one (Your pass log's days). With no audit workflow yet, Run the audit makes it first. Once the audit finishes, its report takes the same three panes as Live, and Copy report link in the title row opens it for anyone signed in to your account. Run another audit, under the report, goes back to the steps.
Recently decided
With no pass chosen, the right pane lists the recommendations decided lately, newest first: when, what each changed and how it was decided (Followed, Declined or Expired), and the pass. It lists the newest 50 and says how far back they go.
A followed one names the bridges its run was delivered to ("run delivered to yamcs") once one has read it. The delivery report counts what a bridge carried out per run, not per decision, so it does not say which recommendation was acted on. For two minutes after you send one, the list reads every few seconds, so that shows within a bridge poll or two.
A pass
A chosen pass fills the right pane, linkable as ?pass=<id>. Escape, or the close in its title row, goes back to Recently decided. It holds the pass's name and when it flew, one sentence on how it ended and where that landed against its forecast ("Failed: 0.0 of 3.0 GB possible, 1.0 GB below its range of 1.0 to 3.2 GB (expected 2.4 GB)."), then the Breakdown: each step of the ledger by name, its value and where the value came from.
In the API
| On screen | Call |
|---|---|
| Upcoming passes and their figures | GET /predictions?workflow=<slug>, narrowed with station_id, antenna_id or satellite_id |
| Price the next 48 hours | POST /predictions with "deliver": false and the window; see Long runs |
| Recommendations waiting, and decided | GET /predictions/actions?workflow=<slug>, with by and at on each decision |
| The past report | GET /predictions?workflow=<slug>, or ?run_id= for a pinned report |
| Review the last 7 days | POST /predictions with the past window and "dataset": "live" |
| An audit | POST /telemetry/pass-logs, then a run with "dataset": "audit" |
The tab reads with a session made from your sign-in; it needs no key of yours. Accuracy explains the past figures, and Data modes and datasets what Live and Audit read.