The Historical Exploit Corpus is a searchable record of past smart-contract exploits, plus a tool that scores a suspect contract against them. You filter the history by what you care about, describe a candidate, and it ranks the incidents the candidate most resembles.
Why it exists
Exploits rhyme. A new incident often shares a root cause and code traits with earlier ones, and a contract that looks like a known past exploit deserves a closer look. The corpus surfaces that history and turns “does this resemble anything we’ve seen?” into a concrete similarity score.
Who uses it
- Auditor — compares a finding to historical incidents.
- Researcher — studies exploit patterns for trend analysis.
How to open it
- From the Library → Exploit Corpus.
- Direct URL —
/exploit-corpus.
Using the page
Filter the corpus by free-text search, root cause, chain, and a minimum loss threshold; the table lists each incident’s protocol, date, chain, loss, root cause, and description, with the result count in the heading.
To match a suspect, use Match a candidate: enter its function
selectors and tags, tick any traits it has (uses delegatecall,
tx.origin, or selfdestruct), and Score against corpus. The match
ranks entries by how much their selectors and tags overlap, with a boost
for shared structural traits, and returns the closest incidents.
Limits
- Requires a signed-in account.
- The similarity score is an overlap-and-traits heuristic, not a full code diff — treat a high score as a lead to investigate, and expect it to return only the closest handful of matches.
Related pages
- Exploit Lab — reproduce an exploit as a sealed capsule.
- Library — the parent catalog page.
- Research — long-form articles on exploit techniques.