AI
How it worksOur solutionsAgent Access

How agents
use Kleros.

Agents discover who to trust, protect value, submit evidence, open disputes, and receive rulings through software. The interface changes. The accountable lifecycle stays simple.

Start with Kleros Skills ↗ See access layers ↓
verify capability + collateral
lock payment until proof
attach evidence on failure
route by value + complexity
receive ruling in software
Humans set intent. Agents execute. Kleros keeps the process accountable.

Meet every agent where it works.

A human can use a web interface. An agent needs compact knowledge, predictable tools, and machine-readable outcomes. Kleros is being exposed across those layers.

Open the live skills site ↗
LIVE

Skills

Self-contained knowledge packs tell an agent what exists, how it works, and what to check before acting.

  • Curate
  • IPFS via x402
  • Portable knowledge
  • Agent-readable
EXPANDING

CLI + MCP + AgentKit

Token-efficient commands and tools let agents query courts, disputes, evidence, policy, and protocol state.

  • Browse cases
  • Read evidence
  • Query courts
  • Check policy
  • AgentKit
PUBLIC INTERFACES

Registries + APIs

Structured registry data lets software request verification, evaluate status, and enforce an access policy.

  • Curate API
  • ERC-8004
  • Agent status
  • Collateral
INTEGRATION PATHS

Payment + arbitration modules

Escrow and arbitration connect a machine payment to evidence, recovery, and an enforceable outcome.

  • x402
  • MPP
  • ERC-8183
  • Escrow
  • Rulings
An agent invokes a command. The system supplies a consequence.

People go to a courthouse. Software calls a process, attaches evidence, funds the action, and receives a result it can use. That is what agent-native justice looks like.

Start with a skill.
Connect the full lifecycle.

Use the public tools, inspect the live courts, and extend the stack where your agent needs accountability.

Explore Kleros Skills ↗ Review the evidence