The chain that
fights back
An Ethereum L2 where AI validators screen every transaction. Anomalies get paused. Exploits get stopped — before the pool drains.
What it is
An Ethereum L2 where every transaction passes through a screening layer before it executes. AI validators, trained on years of exploit patterns, flag anomalies in milliseconds. Suspect transactions pause. The validator set votes. Confirmed exploits are rejected on-chain.
Core principles
Screening at chain speed
Three tiers: deterministic rules on 100% of txs, statistical models on anomalies, LLM review on the hard cases.
Soul-hash commitments
Canonical behavioral profiles committed on-chain as a Merkle root. Different validators, same profile, same result.
Economic accountability
Validators stake AEGIS and ETH; slash for negligence, earn for honest screening. Users pay gas in ETH — no token friction.
How we're building it
Full roadmap →Six phases from specs-on-paper to acquisition-ready. Phase boundaries are movable — the point is to separate what we know from what we've built from what someone is using.
Scoping & design
In flightEvery major call written down. Core specs drafted; decisions tracked per-issue.
Buildable MVP
Next3-validator local testbed screening historical Ethereum data. Tier 1 rules in Rust. Exploit replay reports recall + false-positive rates.
Public testnet
QueuedPublic testnet with Ethereum bridge, team-operated validators, wallet SDK skeleton. First external dev bridges in and catches a flagged tx.
Mainnet + real TVL
QueuedAEGIS token (stake + governance only, gas stays ETH). Native DeFi primitives live. Slashing on, council constituted. Formal audits.
Ecosystem + proof
QueuedWallets and protocols pick Aegis because the numbers prove it out. Public quarterly reports on exploits prevented.
Labs + exit-readiness
QueuedAegis Labs stands up. License split formalized: MIT chain + screener reference, proprietary model weights. Token governance scoped narrowly. Chain keeps running whatever happens to the company.
Help build it
Aegis is pre-alpha. Active work is issue-tagged and open to contribution — builders and autonomous agents both welcome.