Savant.chat

@savantchat

Multi-agent AI smart contract security. Built for deeper vulnerability detection. Trusted by @1inch, @LidoFinance, @GearboxProtocol

linktr.ee/savantchat

Joined March 2023

Replies Reposts Media Articles

Pinned
Savant.chat
@savantchat
Aug 3
New version of SavantChat is LIVE! 2x more powerful, noticeably faster, cleaner UI. What changed underneath: ▸ In closed test runs, the new engine caught up to 2x more vulnerabilities than the best human auditor working the same code ▸ Stronger models across the pipeline, so
1
Savant.chat
@savantchat
1h
Calling all Robinhood Chain devs! $HOOD chain is hot, and the narrative is hotter. But you know how crypto works: attention rotates in weeks. The faster you ship, the bigger the slice of that attention you actually get. That leaves no room for a 3-week audit queue. If you see it
Robinhood Crypto
@RobinhoodCrypto
Sep 2
Robinhood Chain: two months in. 🔷Total DEX volume: $34.6B 🔷Protocol TVL: $1.27B 🔷190+ Stock Tokens live, with $3B+ in cumulative DEX volume 🔷Total transactions: 576M 🔷Total addresses: 12.3M 🔷Perps trading on Lighter: $7.29B in total volume Gratitude to every user and
Savant.chat
@savantchat
Sep 7
Harmony is proposing to sunset its chain, citing "state actors and AI agents." The bug was a signature count that never verified the signatures. It sat in the open, and it reproduces on the cheapest tier we sell. Test yours against submodule and consensus boundaries this week,
Savant.chat
@savantchat
Aug 12
Around 4 billion $ONE appeared out of nothing overnight, roughly 26% of Harmony's supply. About 2.8 billion left in a single transfer and went to exchanges. Price hit an all time low. No contract was exploited and no existing tokens were stolen. The chain accepted messages that
2
Savant.chat
@savantchat
Sep 4
Here is what we shipped in the first week of autumn: ▸ Aggregation by root cause, which is not deduplication. SavantChat attacks a bug from several angles, so 1 real flaw often surfaces as many findings, each a different consequence. We are experimenting with grouping those
Savant.chat
@savantchat
Aug 28
The last week of summer 2026 focused on one thing: fewer false positives, more real findings. ▸ Stability and new models in the auditor. We kept hardening the AI auditor and folded in the latest models, tuned toward better precision. The goal on every run is the same: surface
Savant.chat
@savantchat
Sep 3
Every AI security tool says it is the best. We put the receipts on the site instead. ▸ Top-6 in the Sherlock Symbiotic contest, against dozens of expert human auditors. The first time an AI auditor placed against humans in a live contest ▸ 100% recall on the Crestal contest, 7