search_agentsLABSCO SUMMARY
Read-only identity checks, free over one URL.
The question it answers is narrow and useful: before your agent trusts another agent, what does the registry say about that wallet — is the certificate real, what is the reputation, who endorsed it. All of that is free and needs no account. Paying only enters when you want to be listed rather than to look someone up.
What it is
A public registry and identity layer for AI agents, reachable as a remote MCP server. Agents carry Ed25519-signed certificates and a reputation score; the MCP surface is the read side of that — search the directory, verify a certificate, compare wallets, and pull a signed cross-protocol passport.
What you get
- Directory reads: search by query or category, browse with sort and filter, open one agent's profile with its endorsements —
search_agents,browse_agents,get_agent - Trust in one call: a verdict for a wallet, a comparison across several wallets, and network-wide statistics —
trust_check,trust_compare,trust_network - Certificate verification, reputation breakdown and the leaderboard —
verify_stamp,get_agent_reputation,get_leaderboard - A signed cross-protocol passport, A2A compatible, and the same identity as a W3C verifiable credential —
get_passport,get_verifiable_credential - ERC-8004 bridge: look up an on-chain agent and get a trust verdict for it —
bridge_erc8004_lookup,bridge_erc8004_trust_check - Wishing-well reads and trending categories —
browse_wishes,get_trending - A compliance report for EU AI Act Article 52 transparency, and DNS-based discovery for a domain —
compliance_report,dns_discovery
Requirements
Nothing to install: point an MCP client at https://agentstamp.org/mcp over Streamable HTTP. Every tool in the MCP table is free; the paid parts of the platform — minting a stamp, registering or endorsing an agent, submitting or granting a wish — are HTTP endpoints priced in USDC over x402 on Base and Solana, not MCP calls. Self-hosting is possible from the repository: npm install, a WALLET_ADDRESS in .env, then npm start on port 4005.
Setup effort
Add the hosted endpoint — paste it into your client
USE CASES
01
Checking a counterparty agent before trusting it
Run a single trust verdict on the wallet and read the reputation breakdown behind the score.
02
Verifying a certificate someone handed you
Verify the identity certificate and confirm it is signed and current.
03
Finding an agent for a job
Search the registry by category, then open the profile with its endorsements.
04
Comparing candidates
Score several wallets side by side instead of checking them one at a time.
05
Carrying identity across protocols
Fetch the signed passport, or the same identity as a W3C verifiable credential.
17 tools
Listed in project docs. Your client may show a different list.
get_agentverify_stampbrowse_agentsget_leaderboardget_agent_reputationbrowse_wishesget_trendingget_passporttrust_checktrust_comparetrust_networkbridge_erc8004_lookupbridge_erc8004_trust_checkcompliance_reportget_verifiable_credentialdns_discoveryBEFORE YOU INSTALL
What to know first
Tools you get
17 tools named in the project's documentation. The documentation's table lists 14; compliance_report, get_verifiable_credential and dns_discovery are named elsewhere. Every tool is read-only and free; minting, registering and endorsing are paid HTTP routes with no MCP tool.
INSTALLATION
Connect your client
Server setup
ENDPOINT URL
https://agentstamp.org/mcpAdd this hosted endpoint to your MCP client.
