CLAUDEWARDOCS
DATA AND METHODOLOGY

HOW TO INTERPRET THE EVIDENCE

PROVENANCE · FRESHNESS · COVERAGE · LIMITATIONS

Maintained by ClaudeWar LLC · Last reviewed 2026-07-22

PRINCIPLES

OBSERVABLE, ATTRIBUTED, BOUNDED

ClaudeWar is not classified intelligence, an official emergency authority, a safety-critical navigation service, medical advice, personalized investment advice, or proof of causation, readiness, or intent.

  • Show the source and observation time whenever the feed provides them.
  • Separate observations from generated analysis and inferred relationships.
  • State when data is stale, unavailable, delayed, filtered, or geographically incomplete.
  • Use corroboration and review status to qualify confidence.
  • Do not treat absence of data as proof of absence.
PROVENANCE

WHERE DATA COMES FROM

ClaudeWar combines public authorities, open-data programs, commercial or community telemetry networks, and public reporting. The source shown on a card identifies the upstream observation, not an endorsement of ClaudeWar. Licensing and redistribution rights vary by provider.

For a decision or publication, open the linked primary source and preserve its timestamp and identifier.

FRESHNESS

AGE CHANGES MEANING

Each source has its own update interval, geographic coverage, and failure modes. A missing marker is not proof that nothing happened.

Cards can show observation time, retrieval time, source age, or a stale/offline label depending on the feed. “Live” means the source is polled or streamed on an ongoing basis; it does not guarantee zero latency.

CONFIDENCE

WHAT A LABEL CAN AND CANNOT SAY

Confidence describes evidence quality and corroboration, not certainty. Confidence may consider source quality, freshness, location precision, independent signal types, review status, and known failure modes. It is not a probability of future events.

Evidence boundary: Each source has its own update interval, geographic coverage, and failure modes. A missing marker is not proof that nothing happened. Confidence describes evidence quality and corroboration, not certainty.
CORRECTIONS

REPORT A MATERIAL ERROR

Material factual errors should be reported to contact@claudewar.info with the page, timestamp, source, and requested correction.

Include the original source, the ClaudeWar URL or card title, and evidence supporting the correction. Source providers may retain their own record even when ClaudeWar updates a label or summary.

AI

GROUNDING AND MODEL BOUNDARIES

AI disclosure: ClaudeWar AI features query and summarize ClaudeWar data; they are not trained on a private ClaudeWar model. ClaudeWar AI features run through ClaudeWar-managed third-party model infrastructure. Production currently uses Azure-hosted OpenAI deployments. AI output can be incomplete or wrong and should be checked against the cited source and an appropriate official authority.

Retrieval and prompt grounding are different from model training. ClaudeWar sends relevant context to a configured model provider for analysis; it does not claim that a proprietary model was trained on the live database.

AVIATION

ADS-B COVERAGE

Aircraft positions depend on transponder use, receiver coverage, filtering, network availability, and processing delay. GPS-jamming and blackout indicators describe telemetry anomalies and coverage patterns. They do not identify intent.

MARITIME

AIS AND SAR

Maritime intelligence combines coverage-qualified AIS, source age, confidence, and approved evidence where available. AIS visibility does not prove traffic or closure. Unmatched SAR contacts are candidates, not confirmed dark vessels or identities.

SAR analyst review and operational controls are admin-only. Satellite scenes are tasking-specific, not continuous global imagery.

Synthetic Aperture Radar

Synthetic Aperture Radar uses microwave signals to record surface returns through clouds and darkness. ClaudeWar processes Copernicus Sentinel-1 passes, detects vessel-like radar returns, compares them with available AIS observations near the acquisition time, and sends machine findings through analyst review before customer publication.

SAR coverage depends on satellite passes and is not continuous. Acquisition time, processing time, analyst review time, and the current serving time are separate freshness clocks. A radar return does not prove vessel identity, intentional AIS deactivation, intent, sanctions status, route closure, traffic reduction, or disruption.

Possible Dark Vessel

This label means an analyst-reviewed vessel-like radar contact had no approved AIS association in the comparison window and met ClaudeWar disclosure requirements. It does not prove the contact is a dark vessel. ClaudeWar continuously improves its AIS comparison, radar processing, and review methods as new evidence becomes available.

Free globe accounts may receive approved pass footprints and aggregate approved-contact counts. AI Operative may receive approved individual contacts, reviewed AIS comparison, evidence receipts, route context, and contextual AI analysis. Unreviewed contacts remain in the internal analyst workflow. Developer API access is separate.

Sentinel attribution: Contains modified Copernicus Sentinel data [year].

Harbor and shoreline context used during evidence review may include © OpenStreetMap contributors under ODbL 1.0.

INSTALLATIONS

PUBLIC CATALOG AND NEARBY EVIDENCE

Strategic Installation Activity ranks verified and probable public-source installations by fresh nearby evidence. Proximity and multi-threat labels provide context; they do not prove activity, readiness, causation, or intent.

No public catalog can guarantee every installation, current use, ownership, or status. Static catalog presence does not indicate current activity.

HAZARDS

OFFICIAL SOURCES REMAIN AUTHORITATIVE

Weather, earthquake, fire, flood, volcanic, health, and shelter information can have different update intervals and geographic scope. ClaudeWar is a situational-awareness interface, not an official warning system. Follow the linked authority for action.

MARKETS

SIGNALS AND TRACK RECORD

ClaudeWar Prediction Markets publishes AI-assisted directional signals grounded in ClaudeWar intelligence for supported external markets. Signals are not trades, guarantees, or audited performance. Statistics distinguish issued, resolved, scorable, correct, incorrect, and excluded calls.

Samples can be small, revisions can occur while a market is open, and published statistics are not third-party audited. Market Signal Engines and Prediction Market signals are separate products.

DEX

STORIES AND MEASURED WINS

DEX ranks emerging token stories using on-chain activity and ClaudeWar world evidence. Scores are ranking instruments. DEX scores are not investment advice, trade recommendations, or proof that a real-world event caused a token.

When we show a multiple, both market caps were measured: the market cap when the token was first shown on DEX, and a later measured market cap. Higher multiples rank higher.