Skip to content
neilweitzelPublic

About

Name-free sector research from public incident claims. Evidence kept private, uncertainty kept visible.

Topics

Resources

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Repository files navigation

xevents

Study ransomware claims without amplifying the leak.

xevents turns observed ransomware listing claims into name-free, weekly and monthly sector aggregates for security practitioners and researchers. It preserves the evidence privately and keeps uncertainty visible in the public results. A listing is a claim, not confirmation that a breach occurred.

The aim is to make observed activity useful for research without turning a research tool into another directory of victims or a distribution channel for attacker publicity. xevents collects public listing metadata, groups repeat claims, checks eligibility and cautiously assigns sectors. Only the resulting privacy-checked aggregates become public.

Open the research app · Read the methodology · Explore the documentation

What you can use it for

  • Explore observed activity: compare published claim counts by sector and retrieval week, without a public directory of named victims or threat actors.
  • Build a research snapshot: filter the view and download the selected aggregate data as JSON.
  • Understand the limits: see withheld cells, data freshness and the coverage boundaries alongside the numbers.

This is not a breach registry, an organization risk score, a complete census of ransomware activity or a feed for automated blocking. Unlike xfeeds, xevents is built for interpreting incident claims over time, not distributing indicators for enforcement.

Current status

The public app is available. The unattended release integration is approved for research release-candidate burn-in; it is not yet a graduated production service. End-to-end activation has completed successfully, including automatic publication and verified Pages deployment. The collector is configured to wake every two hours, with at least six hours between successful source captures of up to 200 recent records, and publishes at most one verified release per UTC day (ADR 0036, ADR 0037). A scheduled wake-up is not necessarily a new source capture, and a new capture is not necessarily a new public release.

The app loads published research data by default. If no dataset has been published, it says so instead of substituting sample numbers. An explicitly labeled synthetic demo is available separately and must not be cited as research.

Activity before publishable counts

The app explains work in progress using the latest verified snapshot:

  • Claims assessed privately: cumulative grouped claims, including those not eligible for publication, in bands of 25. “Fewer than 25” includes zero; older snapshots without this measure say “Not reported.”
  • Claims in published counts: the sum of numeric sector-week cells, not a total of all private claims or confirmed incidents.
  • Published sector-week counts: how many cells contain publishable numbers.
  • Published sector-month counts: how many monthly rollup cells contain publishable numbers. Rollups are never added to the published total.
  • Latest source capture: when the source was last successfully captured for this snapshot, not when the page happened to deploy.

Zero published counts can coexist with ongoing private assessment. Exact small private totals and per-gate rejection counts are not exposed. These are snapshot measures, not a live operational-health dashboard or an approval rate.

How to read the results

  • Claims, not confirmed incidents: the initial source is RansomLook. Repeated observations do not provide independent corroboration.
  • Retrieval time, not attack time: reporting weeks start on Monday in UTC, based on when xevents first retrieved a claim. A month holds the weeks that begin in it.
  • Partial coverage: the initial collector samples a bounded recent-record window. It cannot establish how many claims were missed or whether a listing was later removed.
  • Conservative classification: the RC uses description-based sector heuristics. Ambiguous descriptions remain unclassified rather than acquiring an invented sector or confidence score.
  • Small cells stay withheld: counts below five, including zero, are shown as withheld. Missing and withheld values are not evidence of no activity.

The research guide explains appropriate comparisons, exports, privacy limitations and the difference between a claim and an incident.

How it works

Public listing metadata
        |
Private collection and evidence preservation
        |
Automatic eligibility, sector aggregation and exact-output privacy checks
        |
Signed release, protected merge and verified deployment
        |
Public research app and aggregate exports

The automated release path is active for RC burn-in. Routine eligible records do not require individual human approval. Privacy exceptions remain withheld; failed integrity or publication checks stop that release and preserve the previous site. The browser reads public aggregates only and has no access to private evidence or credentials.

Raw source text, organization and threat-actor names, screenshots and record-level evidence stay private. The public repository contains the app, aggregate export surface, verification code and the methodology needed to examine the design. Public export hashes verify published bytes; they are not record-level disclosures of private evidence.

Go deeper

For a non-sensitive aggregate or methodology correction, use the research correction form. Use GitHub private reporting for privacy, identity, credential or sensitive-evidence concerns.

Attribution

The initial derived source is RansomLook. RansomLook identifies its website, API and datasets as CC BY 4.0. xevents transforms observations into restricted aggregate outputs; its classifications and limitations are its own. Source licensing does not make sensitive personal information safe to publish.

Every source's terms, the reading xevents takes of them, and what that source may do (publish, corroborate, enrich or nothing) are recorded in config/source-registry.json, with verbatim terms in research/licence-research-2026-09.md and the policy in ADR 0030. Terms are read as written. Where a publisher states no licence, that is recorded as silence, not as a grant.

In short: xevents follows every source licence as we read it, and takes real care to do so. Mistakes can still happen. If you think we have read your terms wrong, report a bug or open a pull request and it will be corrected immediately. Sources are removed or restricted on request.

About

Name-free sector research from public incident claims. Evidence kept private, uncertainty kept visible.

Topics

Resources

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Releases

Packages

Contributors

Languages