I’m Panther. I review Move systems on Sui, Aptos, and Movement — including prediction markets — plus complex DeFi, backed by production work across Rust/Solana, EVM, Cairo, and Go for protocol teams and ten audit firms.
Code reviewed at the protocols other protocols depend on.
// ten of the twenty-four · section 03 has the full record, severity counts, links
Aave
Lido
Optimism
OpenZeppelin
Polygon
ZKsync
Balancer v3
Sui Framework
DeepBook
Starknet
01 / disclosureCVE-2024-45304
A published advisory in OpenZeppelin's Cairo library
A stale pending owner survived renounce_ownership and could still claim a contract its users believed was ownerless. OpenZeppelin fixed the state cleanup in v0.16.0.
One accepted High and one accepted Medium on the Aptos Move port — configuration written to one address, read from another, aborting every core getter.
An immutability invariant proved on a Sui Move lender
Modelled and machine-checked with Certora's Sui Prover: an obligation's eMode assignment cannot change mid-position, verified across every target function plus sanity checks.
OpenZeppelin Cairo — the pending owner that survived renunciation
Context
Review of OpenZeppelin's Cairo OwnableTwoStep state lifecycle.
Panther's role
Independent researcher; privately disclosed the issue to OpenZeppelin.
Risk & outcome
A stale pending owner could accept ownership after renunciation, restoring admin control to a contract users believed was ownerless. OpenZeppelin fixed the state cleanup in v0.16.0.
Evidence
Published as CVE-2024-45304 / GHSA-w2px-25pm-2cf9.
Sui lending — proving an eMode assignment cannot change mid-position
Context
A real Sui Move lending engagement delivered firm-side through Sherlock; sensitive protocol details remain private.
Panther's role
Security reviewer and formal-verification contributor using Certora's Sui Prover.
Risk & outcome
If an obligation could switch eMode groups, collateral and borrowing rules could cross risk tiers. The modeled immutability invariant passed across six target functions plus sanity checks.
Evidence
The report remains private; the property, methodology, limits, and verification result are documented publicly.
Grouped by language, ordered by weight of record. Bar length is portfolio-record count — not self-assessed skill. Every named codebase links to its entry on the public work record; nothing here is an endorsement.
24 of 76 records shown · the rest include private firm-side engagements under NDAopen the full work record →
#04 engagement_lanes
One specialty. Two ways to engage it.
// one specialty, adapted to protocol teams and audit firms
01for protocol teams
Direct protocol security review
Focused and comprehensive reviews for Move on Sui, Aptos and Movement, Solana Rust, and complex EVM DeFi, from threat model through remediation review.
Move-specific ownership, capability, object, resource, and PTB analysis
Accounting, solvency, oracle, liquidation, and cross-chain invariants
Prediction-market state machines: pricing, signed oracles, settlement, NAV, LP queues, and PTB composition
Report-ready findings with exploit paths and actionable mitigations
Formal verification supportinvariant design, Certora CVL, Sui and Move prover workflows
#05 review_protocol
I follow state, authority, and value—not just functions.
// state transitions over checklist scanning; evidence over intuition
01
invariant model
Model the promise
Turn documentation, trust boundaries, privileged roles, and protocol economics into explicit properties the system must preserve.
02
authority + asset flow
Trace the whole system
Follow capabilities, value, oracle data, and cross-module state through the real execution paths—including composed transactions.
03
break + prove
Challenge, reproduce, close
Break the model at its boundaries, reduce material issues to reviewable evidence, then verify the fix against the original invariant.
#06 delivered_through
Ten firms have put my name on their reports.
Counts are entries on the public work record. A further 34 engagements sit in the NDA-safe private ledger, with client names shown only where disclosure permits.
Pashov Audit Group14 public
Zenith8 public
Adevar Labs6 public
Sherlock4 public
Cantina3 public
Three Sigma3 public
Zellic2 public
Cyfrin1 public
Accretion1 public
BurraSecNDA ledger
“The auditor bonus goes to @theblackpantherhere for this one! Great performance by all others as well, thank you!”
Pashov / Founder, Pashov Audit Group repeat firm-side reviews
“Just wanted to drop a positive feedback, I really like both your skills and dedication. It's great working with you ser.”
Nic / Security Audit Lead, Three Sigma firm-side review work
“I wanted to thank you for your insanely good work throughout the engagement, you submitted a lot of findings, and each one was of high quality. I would recommend you eyes closed.”
Salah Ismail / Security Researcher, Adevar Labs firm-side engagement
Move systems on Sui, Aptos, or Movement, and complex DeFi—especially lending, perps, DEX/CLOB, vault, staking, RWA, oracle, and cross-chain logic. Solana Rust and EVM Solidity scopes are regular work too; other ecosystems are considered when the protocol mechanics match the review depth.
How are timeline and pricing determined?
Per scope and complexity. Focused reviews can fit inside a week; comprehensive reviews commonly run one to three weeks. Share a scope summary, estimated LoC, repository or documentation, and target dates for a concrete quote.
Can the engagement stay private?
Yes. Most firm-side work is under NDA. A mutual NDA can be established before private code access, and public disclosure happens only where the engagement permits it.
Can audit firms reserve reviewer capacity?
Yes. Use the firm-side intake route and include ecosystem, approximate scope, delivery window, expected role, and your internal reporting workflow.
Give the protocol an adversarial review before users do.
Send the ecosystem, scope or estimated LoC, repository/docs, target dates, and the outcome you need. Start with a public link or scope summary; private repository access can move to an NDA-backed channel.