Governance for your AI agents
Control what your agent can do. Verify what it did. Show the proof to anyone who asks.
Aegis gives operators on-chain policy enforcement and a full audit trail — so an agent can go from demo to production without giving up accountability.
The three jobs Aegis does for every agent
An agent isn't safe to deploy until you can answer all three questions: what is it allowed to do, did it stay on purpose, and what actually happened.
Control
The agent can't do anything you didn't authorize.
- • Spend, frequency, and asset rules — enforced before a transaction lands
- • Time-bounded authority, so the agent only operates in the window you grant
- • Approved actions, tokens, protocols, and chains — nothing outside the list
Judgment
The agent's actions match its stated purpose.
- • Each action ties to a declared purpose, checked against the transaction
- • A pass/fail decision on submitted work before any funds move
- • Compromised keys still can't exceed policy — the contract is the final word
Visibility
You know what happened and can prove it.
- • Every authorization, action, and outcome captured in one record
- • Behavioral signals surfaced as the agent runs — not after the fact
- • Compliance-ready exports for the questions that come later
What changes when you run on Aegis
On-chain enforcement
Policies are enforced by the chain itself. Disallowed transactions never execute.
Isolated bot signers
Each agent gets its own keypair, completely separate from your personal wallet.
Full audit trail
Every action logged with on-chain proof. Filterable, exportable, defensible.
Framework-agnostic
Works with LangChain, Claude, OpenAI, Crew AI, or any agent code. REST API.
Built for agentic payments — x402 included
Agents increasingly pay per request over x402 — gasless stablecoin authorizations (EIP-3009) that settle without a normal transaction. Most guardrail stacks never see these payments, so they slip past every spending limit.
- Every x402 payment is checked against your policy before it settles — over-limit payments are refused.
- x402 spend counts toward the same daily volume and per-transaction caps as everything else — no blind spot.
- Each payment is captured in the audit trail, with on-chain proof and exportable records.
Where Aegis fits
Trading & DeFi agents
Autonomous trading, liquidity, and rebalancing — operating only on approved venues, within caps you set per transaction and per day.
Treasury operations
Agents that move funds inside an organization — payments, swaps, rebalances — with constraints that match policy and an audit trail compliance can use.
Orchestration agents
Coordinate borrow, swap, and stake flows across protocols — without ability to deviate from approved sequences.
Enterprise automation
Business workflows where every action needs a policy reference and a receipt — payments, transfers, accounting reconciliation.
Getting started
Sign in with your wallet
No email, no password. Connect with SIWE and you're in.
Register an agent
Generate a dedicated signer so the agent never touches your personal wallet. Free during early access.
Write a policy
Define what the agent can do — actions, tokens, protocols, spending limits, time windows.
Grant permission and run
Bind the agent to the policy. Disallowed actions can't execute, period.
Watch the audit log
Every action recorded with on-chain proof. Update policies whenever you need to.
Frequently asked
What if my agent gets compromised?
How do I update a policy?
Does this work with my LLM?
Is it ready for production?
What networks are supported?
How much does it cost?
Looking for technical depth — contract addresses, API reference, integration examples? Read the docs.
Take an agent to production
Define what it can do. Verify what it did. Keep the proof. Free during early access.