Web3 Hacks Dataset

Explore a public sample from SCH's incident database, then request a secure CSV download link for exploit analysis, quarterly incident research, security content, and attack-class trend studies.

  • No public API
  • Signed CSV link
  • Source-linked rows

Sample records preview

A 8-row sample of the dataset shape. Request the full CSV for the complete file.

Explore dashboard
Project Date Loss Chain Attack class
Aztec Bridge Unclassified
Taiko Other
JaredFromSubway.eth MEV Bot Other
Aztec Other
Aztec Connect Unclassified
Humanity Protocol Unclassified
Thetanuts Finance Flash Loans Attacks
Aztec Other

Why this dataset exists

Most Web3 incident roundups are readable, but difficult to reuse in analysis. This dataset turns SCH's incident tracking into a researcher-friendly export with consistent attack-class mapping, source URLs, and structured loss data that can be sorted, filtered, and cited directly.

Included fields

The export is designed for analysts, researchers, and content teams who need reusable structured data rather than screenshots.

slugStable incident identifierUseful for citation and direct linking.
dateISO incident dateSupports quarterly and yearly trend analysis.
amount_lost_usd_numericParsed USD lossEnables sorting and aggregation without reparsing display strings.
chainsAffected chain groupSupports cross-chain incident comparison.
techniqueSource technique labelPreserves provider framing when incidents are hard to normalize.
attack_classesSCH attack taxonomyConnects incident analysis to exploit education and prevention categories.
auditorsAudit attributionUseful for ecosystem and firm-level studies where available.
poc_urlsProof-of-concept linksShortens the path from incident data to reproduction material.
detail_urlCanonical SCH incident pageProvides a citeable deep-link for context.

Methodology and usage notes

Rows are drawn from the same incident database that powers the SCH Web3 Hacks Dashboard. Attack-class labels are mapped against SCH's public exploit taxonomy. When public losses are missing or unparseable, the numeric loss field remains empty rather than being guessed.

Source links stay attached so researchers can inspect the original record before treating any summary as final. Use the interactive dashboard when you want visual filtering.