Assistant
Use the assistant to inspect the loaded fleet, calculate statistics, predict link SNR, evaluate service availability, and control the globe. Ask in plain language; the assistant chooses tools and can combine several calls to complete a request. The tool names below identify console capabilities, not public HTTP endpoints.
Guest access includes limited assistant queries. Prediction tools require an enabled backend.
Ask a question
Name the metric, assets, and time window. Select an asset on the globe or type @ to add context. Remove context pills when they no longer apply.
- "What is the mean SNR for the current fleet?"
- "Show mean utilization by ground station over the last hour."
- "Predict SNR for this satellite-to-gateway link."
- "How many hours of service can the loaded constellation provide over Tokyo in the next 24 hours?"
- "What happens if the Svalbard gateway goes down?"
The assistant analyzes assets already loaded into the scene. Catalog assets may have orbital elements without telemetry or observed links. This build does not provide assistant tools for creating new satellites, gateways, or constellation designs.
Analysis and prediction tools
| Tool | What it does | Example request |
|---|---|---|
get_scene_state | Reads the scene clock, playback window, asset counts, and source breakdown. | "What fleet and time window are loaded?" |
get_fleet_summary | Summarizes fleet health, utilization, latency, capacity, and assets at risk. | "Summarize fleet health." |
get_entity_detail | Inspects a node or link by ID, name, or NORAD ID. Returns available orbital, telemetry, connectivity, or link-budget details. | "Inspect the selected satellite's orbit and links." |
query_fleet | Calculates snapshot statistics, historical series, or grouped comparisons. | "Compare mean link SNR by ground station over the last hour." |
query_chart | Creates a chart from the current snapshot, with a metric, grouping, sorting, and chart type. | "Chart current utilization by ground station." |
render_fleet_chart | Creates fleet percentile, histogram, or latency-trend charts. | "Plot fleet P90 link latency over time." |
predict_link_snr | Requests SNR inference for a source/target pair, subject to model access and available inputs. | "Predict SNR between this satellite and this gateway." |
analyze_service_availability | Evaluates site coverage, service hours, contact passes, gaps, and modeled link performance over a window. | "Show coverage and the longest service gap over Tokyo for the next 24 hours." |
Fleet statistics and charts
For query_fleet, choose a property such as link SNR, latency, utilization, throughput, capacity, reliability, node queue depth, altitude, or rain rate. Supported statistics include mean, median, min, max, sum, count, standard deviation, percentiles, outliers, and confidence summaries.
Use shape=point for the current snapshot, series for a trend, or grouped for a comparison. Grouping dimensions include ground station, satellite, link type, spectrum, node type, health, orbit band, and region. A window_minutes request uses a trailing window ending at the selected replay cursor. Results depend on the loaded history and report their sample coverage and exclusions.
query_chart supports bar, horizontal bar, stacked bar, line, area, scatter, pie, donut, radar, and histogram charts. It reads the current snapshot; choosing a line chart does not turn snapshot data into history. Use a historical query_fleet request or the latency-trend option in render_fleet_chart for time-series analysis.
SNR predictions
predict_link_snr takes source_id and target_id, with an optional quality_tier permitted by your subscription. Identify both endpoints or select the link before asking. Check the returned provenance, model, timestamp, and units. A prediction is distinct from the measured SNR available through query_fleet.
Availability depends on the console build, account access, backend, and recent link inputs. A disabled or unavailable prediction is not a zero-SNR result. See Predictions for input fields, forecast metadata, and per-link errors.
Service availability
analyze_service_availability accepts a known city name or explicit lat and lon. Its main settings are:
| Setting | Default | Meaning |
|---|---|---|
window_hours | 24 | Evaluation window; supported range is 1–48 hours. |
step_seconds | 120 | Sampling interval; minimum 30 seconds. |
min_elevation_deg | 25 | User-terminal elevation mask. |
rain_rate_mm_hr | 0 | Uniform rain assumption for modeled link performance. |
Check the resolved coordinates, eligible satellites, excluded assets, and evaluated window. Coverage and estimated throughput describe the model under those assumptions; they do not establish measured service or reserved capacity. Assets need usable scene tracks or orbital data for the requested window.
Scenarios and imaging
| Tool | What it does | Example request |
|---|---|---|
simulate_gateway_outage | Applies a gateway outage by name-resolved IDs, country, region, geographic bounds, or the busiest measured gateway. Reports connectivity changes. | "Simulate an outage at the busiest gateway." |
list_active_mutations | Lists changes applied to the active scenario. | "Which outages are currently applied?" |
reset_scenario | Clears scenario mutations and returns analysis to the unmodified fleet. | "Reset the scenario." |
run_mission | Dispatches a simulated imaging-and-inference mission for a named target, with standard, heavy, or maximum workload. | "Image the Panama Canal with a heavy workload." |
Outage scenarios can disable or remove gateways. Check the matched gateway set before interpreting the result. Rehomed or surviving connections do not prove sufficient spare capacity or a solved redistribution of traffic. Reset the scenario before comparing with the original fleet.
Imaging requires a scene with Earth-observation, communications, compute, and ground capabilities. The mission animates tasking, capture, orbital inference, and delivery. See Demo mode for the walkthrough, workload options, and imagery provenance.
Globe and timeline controls
| Tool | What it controls | Example request |
|---|---|---|
select_node | Selects an asset by ID, label, or NORAD ID. | "Select this satellite." |
select_link | Selects a link using its two endpoints. | "Select the link between these assets." |
clear_selection | Clears the selected asset or link. | "Clear my selection." |
set_asset_visibility | Shows, hides, or isolates registry assets by ID, name, or group. | "Show only my ground stations." |
set_filter | Filters by orbital or catalog attributes, such as altitude, inclination, orbit class, or object type. | "Show LEO payloads below 600 km." |
set_constellations | Shows, hides, or isolates catalog groups, optionally limited to named members. | "Show only Iridium." |
clear_globe | Hides assets and overlays and clears display selection/filter state. Does not delete assets or reset scenarios. | "Clear the globe." |
set_camera | Frames the Earth, a coordinate, or a geographic bounding box. | "Center the globe on Tokyo." |
set_layers | Changes links, labels, sky glow, Earth style, and 2D/3D view. | "Switch to 2D and show labels." |
seek_time | Moves to a UTC timestamp or elapsed time within the loaded window. | "Jump to 30 minutes after the replay starts." |
set_playback | Pauses, resumes, or changes playback speed. | "Pause playback." |
set_timeline_mode | Loads demo, replay, or live data when available. | "Load the demo fleet." |
open_panel | Opens search, gallery, assets, settings, or tips, or closes panels. | "Open the chart gallery." |
"Show only" replaces what is visible; "show" adds and "hide" removes visibility. Display changes do not delete fleet assets. Read the result for unmatched names or filters that matched nothing.
Integration tool
get_integration_bundle returns integration information as JSON, an IDE rule, or installation instructions. It requires a token created in Settings → API & integrations and is unavailable to guests. Ask "Show my fleet agent integration instructions." See Fleet SDK and tooling for installation and verification.
Read and save the result
Tool steps show progress during a turn. Stop cancels further work; it does not undo scene changes already applied. Use reset_scenario to clear what-if mutations.
Check an analytical result's data source, time window, asset scope, units, and exclusions. Missing samples, unmatched assets, and failed tools are conditions to resolve, not analytical findings. A dispatched mission has not yet completed or delivered its result.
Capture an analytical result into a report, or save the query for reuse. Saved work is stored in your browser profile; export reports you need elsewhere.