group Platform

wifi Live Connections — ephemeral, in-memory only

trending_up Traffic & Visitors — no IPs, no cookies, resets on restart

visibility Visitor Analytics — persistent, no PII, Redis-backed

security Intelligence

api API Usage — today's request counts

health_and_safety Source Health — pipeline source status at a glance

add_circle Add Keyword

list Active Keywords — used to filter social media posts for relevance

grid_view Keyword Matrix — compound keyword combos (method x action x context) across languages

monitoring All Sources — built-in + custom, with tier status and error tracking

add_circle Custom Feeds — add your own RSS/Atom feeds

add_circle Add Channel / Subreddit / Hashtag

Manage users, assign tiers and roles. Role and tier changes save immediately on dropdown change.

tune Feature Permissions — toggle features per tier (saves immediately)

list Submitted Reports — tips and reports submitted by users via the Report Event form

analytics Detection Stats — injection attempts stripped before LLM processing

history Recent Alerts — in-memory, resets on restart

filter_list Filter — tips flagged or rejected by content screening

Auto-discovered sources, underperforming channels, and broken feeds.

Keywords discovered by the LLM from event analysis. Approving adds them to the OSINT keyword filter.

Hashtags extracted from social media posts. Approving adds them as Mastodon monitoring sources.

AI-generated intelligence summaries per country.

Active refresh tokens — each represents a logged-in session. Sessions expire after 7 days of inactivity.

hexagon Event Types

Click any card to change its colour or icon. Changes apply to the live map immediately.

bar_chart Severity Levels

Click to change severity colours. Counts from last 24 hours.

flight Aircraft Types

Click to change colour. Toggle visibility on the map.

sailing Vessel Types

Click to change colour. Toggle visibility on the map.

layers Map Layers

Base map layers and overlay settings. Hex grid colours follow severity/category settings above.

hexagon Hex Grid Blades

The hex grid clusters events by area. Each blade inherits its colour from the most severe event in the cell. Click to change the blend mode.

Every privileged action — tier changes, org mints/cancels, broadcaster approvals, credit grants — appends a row here. Read-only.

Self-serve checkout for Sentinel Team / Enterprise is intentionally disabled. Buyers come in via the contact form; once qualified, mint the org here. The owner you assign gets effective tier sentinel_team or enterprise on next /me. Seats are enforced when members accept invites.

Broadcast tier subscribers require admin approval before their channels go live.

Messages submitted via the contact form. Stored in memory only — no database, no logs.

500-neuron spiking neural network watching the pipeline. Modulators steer ingestion behaviour; concept + cascade activity surface developing patterns. Cross-check against the heuristic detectors below — they should agree when something real is happening.

bolt Burst Detection

Unusually high event activity for a specific country and event type. A burst means we're seeing significantly more events than normal in the last hour compared to the 7-day average. Higher z-scores indicate stronger anomalies.

speed Entity Velocity

Locations, organisations, and people that are suddenly being mentioned far more than usual. If an entity's velocity spikes, something is developing around it — even if individual events haven't been classified as critical yet.

location_on Geographic Hotspots

Geographic areas (~55km grid) where event density is significantly above normal. These are places where something is happening right now that's generating more intelligence reports than we'd expect. Recent events from the area are shown for context.

trending_up Emerging Topics

Clusters of similar events detected by machine learning. When multiple events start converging around a common theme, they appear here. "Emerging" means the pattern is new; "Growing" means an existing pattern is intensifying.

fact_check Prediction Accuracy

Tracks whether our causal chain predictions come true. Every prediction is logged with a deadline based on the expected lag time. The system checks hourly for matching events and updates accuracy scores per chain. Chains with higher accuracy should be weighted more heavily in future assessments.

timeline Recent Predictions

The last 50 predictions made by the system, with their current status. Pending predictions are within their lag window and waiting for confirmation. Confirmed means a matching event was found. Expired means the lag window passed without a match.

Rolling reliability scores based on corroboration rate, disinfo events, and rejection rate. Beta-distribution posterior (Bayesian). Sources with fewer than 10 events score 0.50 (neutral prior).