Sigvex
Operations3 min read

Historical Exploit Corpus

Searchable library of historical smart-contract exploits with cause and chain filters and a candidate-match tool.

Open Historical Exploit CorpusAll guides
corpushistoricalexploits

A catalog of historical exploit entries with filters for Search, Root cause, Chain, and Min loss (USD); a Match a candidate panel with Selectors, Tags, and trait checkboxes (uses DELEGATECALL, uses tx.origin, uses SELFDESTRUCT); and a Score against corpus action that ranks entries by Jaccard + structural-boost similarity.

Why it exists

Many exploits share root causes and code traits with earlier incidents. The corpus surfaces that history and provides a candidate-match tool so a reviewer can check whether a suspect contract resembles a known past exploit.

Who uses it

  • Auditor — compares a finding to historical incidents.
  • Researcher — studies exploit patterns for trend analysis.

How to open it

  1. Sidebar → Library → Exploit Corpus.
  2. /library → Exploit Corpus card.
  3. Direct URL — /exploit-corpus.

Using the page

  1. Search — free-text filter over protocol, description, and tags.
  2. Root cause — scopes by root-cause category (defaults to Any).
  3. Chain — scopes by chain (defaults to Any).
  4. Min loss (USD) — numeric threshold filter.
  5. Match a candidate — enter Selectors (comma-separated, 0x-prefixed) and Tags (comma-separated), and tick any trait boxes (uses DELEGATECALL, uses tx.origin, uses SELFDESTRUCT).
  6. Score against corpus — ranks entries by selector + tag Jaccard with a structural boost for matching traits.
  7. Corpus — table with ID, Protocol, Date, Chain, Loss, Root cause, and Description columns; the heading shows the current result count.

Inputs and outputs

Input Source Required
Search Text input no
Root cause Dropdown no
Chain Dropdown no
Min loss (USD) Number input no
Candidate selectors Text input for Score against corpus
Candidate tags Text input for Score against corpus
Candidate traits Checkboxes no
Output Format Destination
Ranked matches In-page HTML
Filtered corpus In-page HTML

Controls reference

  • Search — free-text filter.
  • Root cause — root-cause scoping.
  • Chain — chain scoping.
  • Min loss (USD) — minimum-loss threshold.
  • Selectors / Tags — candidate inputs for similarity scoring.
  • Candidate trait checkboxes — traits (DELEGATECALL, tx.origin, SELFDESTRUCT) that add structural boost.
  • Score against corpus — runs the similarity match.

Limits

  • Requires a signed-in account.
  • The corpus is served by the API; filters re-fetch from the server.
  • Similarity scoring uses Jaccard over selectors + tags plus a structural boost — it is not a full code diff.
  • Matches are capped at the top 8 entries scoring ≥ 0.05.

Troubleshooting

Symptom or error Cause Action
Empty list Filters exclude every entry Clear Search, reset Root cause/Chain to Any, and set Min loss to 0.
Score against corpus ranks nothing No selectors/tags supplied or none matched above 0.05 Add more selectors or tags.
  • Exploit Lab — reproduce exploits via capsules.
  • Library — parent catalog page.
  • Research — long-form articles covering exploit techniques.