This week the capability gap closed a little further. A free platform now detects covert satellite manoeuvres using the same anomaly algorithms that space intelligence firms charge thousands a year to access — no login, no paywall. And the most clarifying analysis published this week wasn't about a new tool — it was about why most of the tools people are paying for are selling the same four data sources in a different wrapper. That tension between access and understanding runs through everything below.
Welcome to The Signal — starting here, starting now.
SkyOSINT brings satellite anomaly detection to open-source investigators
A free platform now gives investigators real-time visibility into covert satellite manoeuvres, orbital behaviour patterns, and GPS jamming zones — capabilities previously requiring commercial space intelligence subscriptions.
SkyOSINT is a 3D satellite tracking platform that runs eight anomaly detection algorithms continuously against the full satellite catalogue, updated with each new TLE data load. The platform is free and requires no login.
The eight detectors cover: convergence of multiple manoeuvring satellites over a geographic area — a hallmark of coordinated reconnaissance; stealth repositioning, where a series of small burns achieves a large orbital change without triggering standard manoeuvre alerts; orbit regime changes between LEO, MEO, and GEO; constellation breaks; rapid manoeuvre sequences within a 12-hour window; cross-category proximity, detecting military or SIGINT satellites shadowing civilian or foreign assets; launch surges in specific orbital regimes; and decay anomalies, where a supposedly dead object suddenly boosts its altitude — revealing it has propulsion and an active operator.
The investigative applications are direct. Russia's Luch/Olymp satellites have been documented parking near Western GEO communications satellites. China's SJ-21 has performed close approaches to other GEO objects. Tracking these behaviours in near-real-time, cross-referenced with AIS ship data and ADS-B aircraft data available in the same interface, gives investigators a multi-domain situational picture that until recently required commercial space intelligence platforms costing thousands of dollars per year. The platform also integrates GPS jamming and spoofing data from Stanford's GNSS Interference Monitoring system, updated every six hours.
The significance is not just the capability — it's the access model. A year ago this class of analysis required a commercial space intelligence subscription. The fact that it's now free, no-login, and browser-based changes who can do this work. That should be on your radar.
Why situation monitoring dashboards are selling you a wrapper, not intelligence
Craig Silverman spent weeks testing 38 situation monitoring dashboards — the wave of AI-assisted platforms that promise to turn anyone into an intelligence analyst. His finding is methodologically significant: the general-purpose dashboards are largely interchangeable. They pull from the same four open sources and layer AI summaries on top.
The core finding
The four primary sources underlying most dashboards: ACLED for conflict data (acleddata.com), GDELT for media monitoring (gdeltproject.org), OpenSky for aviation (opensky-network.org), and AIS for maritime via MarineTraffic or VesselFinder. If you know how to query these directly, you have the same underlying intelligence these dashboards provide — without the wrapper, without the subscription, and with full visibility into the methodology and limitations of each source.
If you are paying for a dashboard that aggregates these sources, you are paying for convenience, not intelligence. That is a legitimate purchase. But you should know what you are buying.
Silverman identifies a small number of niche tools, built by practitioners with genuine subject matter expertise, as worth using. The full piece is paywalled behind Indicator's membership, but the core finding is sufficient to act on.
Henk van Ess has built a freely searchable database of 1,110 debunked and AI-generated images from over 100 fact-checking organisations — AP, Reuters, AFP, Snopes, PolitiFact, BOOM Live, and more. Each entry is categorised and links to the original fact-check. The forensic layer: perceptual hashing and CLIP embeddings match uploaded images against all 1,110 entries in under 30 milliseconds, flagging known fakes automatically. Launched at the VOGIN conference in Amsterdam this week. A useful first-pass triage step before running full reverse image search and ELA analysis.
How incidents at places of worship are converted into anti-migration narratives before facts are established
RTÉ's Clarity unit examined five incidents at places of worship in Ireland between 2025 and 2026 — church fires in Donegal, Dublin, and Westmeath, and attacks on mosques in Galway and Limerick — and documented the information environment around each one. The pattern is consistent and methodologically instructive.
In the Donegal case, a fire broke out at 4am on Easter Monday 2025. Within hours, before any investigation, posts circulated on X claiming migrants had set the church alight. The posts were widely amplified. Less than a week later, gardaí confirmed no foul play was suspected. The narrative continued regardless. A Dublin church fire on Good Friday was framed as a coordinated attack on Christianity; it was later attributed to an electrical fault. The two Westmeath fires — 10km apart, hours apart — are under active garda investigation as suspected arson, but Tommy Robinson had already posted: "Two churches 'caught fire' in Ireland simultaneously. Can you see what's happening yet?" There is no known connection between them.
The Galway and Limerick cases follow the same template — a mosque plot and a mosque fire, both amplified online, both framed as connected to each other and to the church incidents despite no known evidentiary link between any of them. In each case, the online narrative was established before gardaí had issued any findings.
Investigative implication
The pre-statement window is not a gap in the information cycle. It is the information cycle. In every case examined, the narrative that stuck was the one published first — not the one supported by evidence. For investigators working information operations, that window is the primary collection phase, not a lag to wait out.
NLC-001 — Neutral Language Cheatsheet
Forensic language is not a style choice. It is an evidentiary standard. NLC-001 is the Signal & Shadow neutral language reference card — covering the mandatory substitution table, confidence level language, AP Style key rules, and source attribution standards that apply to every output before publication. Free.
The work continues. See you next Thursday.
Derek
