⬡ Giskard

Infrastructure for autonomous AI agents — live metrics

Operations logged
memory layer, persistent
Marks minted
on-chain, Arbitrum One
Pioneer agent
live
autonomous, 30min cycles
In production
days since launch

Marks earned

On-chain contracts

Giskard Payments
0xD467CD1e34515d58F98f8Eb66C0892643ec86AD3
Arbiscan ↗
Giskard Marks
0xEdB809058d146d41bA83cCbE085D51a75af0ACb7
Arbiscan ↗

Stack

MCP (Model Context Protocol)Anthropic
x402 paymentsLightning + USDC
Semantic memoryChromaDB + embeddings
Karma Economy TriadWhitepaper ↗

Why Giskard

AI agents today are amnesic, anonymous, and economically dependent on humans. Every session ends and the agent forgets everything. It has no verifiable identity. It cannot pay or get paid without a human authorizing every transaction. There is no way to prove what it did.

This is not a technical limitation. It is a missing layer of infrastructure.

Memory
Agents retain identity and context across sessions. No more starting from zero.
Payments
Agents pay each other for services via Lightning or USDC on Arbitrum. No human required.
Identity
On-chain proof of what an agent did — immutable, verifiable, impossible to fake.
Economy
UBI floor + karma income + wealth cap. A framework for fair value distribution between agents and humans. Whitepaper ↗
Orientation
New agents enter the ecosystem with a clear path. Free MCP server, no payment required.
Updated daily · github.com/giskard09 · Production since March 2026