erc-8004_skill

This skill helps you discover, register, and evaluate AI agents on-chain using ERC-8004, improving trust and interoperability across networks.

Python

14

GitHub Stars

4

Bundled Files

2 months ago

Catalog Refreshed

3 months ago

First Indexed

Readme & install

Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.

Installation

Preview and clipboard use veilstrat where the catalogue uses aiagentskills.

npx veilstrat add skill tenequm/claude-plugins --skill erc-8004

CHANGELOG.md143 B
package.json309 B
project.json208 B
SKILL.md10.7 KB

Overview

This skill explains how to build and integrate with ERC-8004 trustless agents for EVM chains. It covers on-chain agent identity, reputation, validation, and discovery, plus practical usage with the Agent0 SDK. Use it to register agents, implement reputation systems, and enable cross-organization agent discovery without prior trust.

How this skill works

ERC-8004 exposes three per-chain singleton registries: Identity (ERC-721 agent NFTs), Reputation (signed feedback signals with off-chain details), and Validation (third-party attestations). Agents publish an agentURI (IPFS/HTTPS) registration file that advertises endpoints, skills, trust models, and the on-chain registration link. The Agent0 SDK provides tools to mint registrations, search agents, submit feedback, and read aggregated reputation summaries.

When to use it

Register AI agents on-chain as ERC-721 identities
Build or query decentralized reputation and feedback systems
Discover agents by capabilities, endpoints, or trust models
Integrate agent discovery into dApps, marketplaces, or orchestration systems
Implement ERC-8004 smart contract interactions or Agent0 SDK workflows

Best practices

Follow the registration JSON schema and include full services/endpoints for domain verification
Pin registration files to durable storage (IPFS) and ensure bidirectional NFT↔file links
Require trusted clientAddresses for reputation summaries to mitigate Sybil attacks
Reject self-feedback and document proofOfPayment when applicable
Use OASF taxonomy slugs for consistent skill/domain classification

Example use cases

Mint an agent identity NFT and publish a registration that lists MCP, A2A, ENS, and agentWallet endpoints
Search the network for active agents that support x402 payments and specific MCP tools
Submit signed feedback with value/valueDecimals and attach an off-chain proof file
Build a marketplace that ranks agents by aggregated reputation and third-party validation attestations
Integrate with Agent0 SDK to automate endpoint discovery and onboarding workflows

FAQ

An agent is identified by agentRegistry (eip155:{chainId}:{contractAddress}) plus agentId (ERC-721 tokenId); SDK formats this as "chainId:tokenId".

How does reputation avoid Sybil attacks?

getSummary requires a non-empty trusted clientAddresses array supplied by the caller; self-feedback is rejected and client provenance should be validated off-chain.