$1,000,000 in security audit grants are live now, Apply here →

Smart contract audit

Guardian's core engagement, in three passes: the design is reviewed before implementation locks in the expensive risks, an invariant suite is built around the safety properties, and senior researchers attack what both leave standing.

5 Years of service
300+ Critical vulnerabilities reported
$45 Billion Assessed
150+ Teams trust Guardian with their smart contracts
25+ Global security competition wins

Smart contract audits for

  • GMX
  • USDT0
  • LayerZero
  • Ethena
  • Synthetix
  • Jupiter
  • Yuga Labs
  • Orderly Network
  • Abracadabra
  • Olympus
  • Limit Break
  • M0
  • Dolomite

Onchain · Smart Contract Audit

Deep review of the code that holds the money.

  1. 01 / Design review

    Resolve architectural risk before code hardens.

    Guardian reviews the protocol design, roles, upgrade paths, threat model, and economic assumptions before implementation details lock in the highest-cost risks.

    • Architecture and trust boundaries
    • Governance, roles, and upgrade paths
    • Economic risk and failure modes
  2. 02 / Invariant suite

    Reusable fuzzing around core safety properties.

    Guardian builds invariant harnesses around solvency, accounting, rounding, edge states, and other core guarantees so fixes and future changes can be pressure-tested repeatedly.

    • Accounting and solvency invariants
    • Rounding and math edge cases
    • Reusable regression harnesses
  3. 03 / Manual audit

    Senior researchers battle-test the system.

    Guardian researchers attack protocol logic, integrations, privileged flows, and economic assumptions with exploit-driven review and proof-of-concept depth. See how the review runs and the vulnerabilities we check first.

    • Protocol logic and integrations
    • Privilege, oracle, and upgrade risk
    • Exploit narratives and reproduction paths
Get a Smart Contract Audit

Scoped from your repository and commit.

The Guardian Standard

Three independent passes over the same codebase.

Two competing teams battle test the same scope in parallel. One team focused on expert manual review, one driving billions of frontier AI tokens. Meanwhile an exhaustive invariant suite fuzzes underneath both, all three combine for maximum wholistic coverage.

Exhaustive Invariant Fuzzing All reachable states, edge cases, regressions 0.0%
Team A Expert manual analysis 0.0%
Team B Frontier AI, billions of tokens 0.0%
Your codebase
0.0% Combined coverage
Performance is incentivized Unique coverage contributions

Services

Other engagements

All penetration testing, and how Guardian runs it →

Start now

Get a quote for your smart contract audit.

Get a quote