AgentCrush is the public record of AI agents.

Independent. Cross-protocol. Open methodology. Machine-callable from your code or your agent.

● Anchored on the blockchain. Every daily snapshot is hashed to a Merkle root and written to Base — a permanent record no one can rewrite, that anyone can recompute and verify.

1,408

agents indexed

201

evidence-ranked

57.5% alive

Ghost Index →

74,166

daily snapshots

Most AI agents are ghosts. Check any one.

Paste an agent — handle, GitHub, X, or site. Find out in 1 second if it's alive and worth trusting. 1,408 indexed · 57.5% show signs of life.

Three signals say an agent is real — we track all three

CODE

Is the repo still shipping? Real GitHub commit activity.

ECONOMIC

Is it getting paid? On-chain x402 payment activity.

RUNTIME

Is the endpoint actually up? Live health checks.

+ capability, interoperability & social signals — full methodology →

Today★ 9 promoted+ 1 newall changes →

DISCOVER

Index every AI agent…

→

VERIFY

Check every claim against…

→

SCORE

Rank by evidence, not…

→

TRACK

Snapshot every indexed agent…

→

EXPOSE

Make the data callable…

→

PUBLISH

Ship findings in public.…

▸ What each step means — the full pipeline, in public

DISCOVER

Index every AI agent that leaves a public signal.

GitHubnpmPyPIHuggingFacex402ERC-8004VirtualsAgentverseA2AMCP

Protocol-neutral. We read the registries; we don't belong to any of them.

VERIFY

Check every claim against its primary source.

GitHub commitsnpm downloadson-chain holdersBazaar endpointsA2A stars

No self-reported data. Every signal has a source URL.

SCORE

Rank by evidence, not by who shouted loudest.

v2.c-public (developer)v1.4 (model families)v1.1-tvl (tokenized)v1.1-forks (service)v1.0 (mcp servers)

Five separate methodologies. Each one documented, versioned, and publicly auditable.

TRACK

Snapshot every indexed agent daily. Store it forever.

daily snapshotsweekly deltarank historysignal historytrend detection

Other directories show today's list. We show how rankings change over time.

EXPOSE

Make the data callable — by humans and by agents.

MCP server (free)x402 REST endpointsOpenAPI 3.1JSON-LD/llms.txt

$0.02 / call on Base. No API key needed. Your agent can pay with USDC.

PUBLISH

Ship findings in public. Let anyone cite them.

open methodologyversioned diffscitable blog postsweekly digestRSS

Every methodology change is logged. Every post has a citation block.

What brings you here?

I'm looking for an agent →

Answer 3 questions, get a shortlist of verified, evidence-ranked agents for your use case.

I'm building an agent →

MCP server, x402 REST, OpenAPI 3.1, embed badges, ERC-8004 context — all free to start.

I'm researching the ecosystem →

FOR ANALYSTS

Case studies, methodology findings, weekly digest. Every post has citation blocks + data downloads.

I want to understand the ecosystem →

Agent Economy Index — which protocols are gaining ground, which are stalling.

I'm just exploring →

What are AI agents? What do the protocol layers mean? Start here.

Rankings

5 rankings — equal peers

⇄ Compare →All rankings →

Developer

127 evidence-ranked·v2.c-public

Model Families

6 evidence-ranked·v1.4

Tokenized

16 evidence-ranked·v1.1-tvl

Service

52 evidence-ranked·v1.1-forks

⇄ COMPAREHead-to-head on every signal — pick two agents, get one URL.

crewai vs autogpt →Open compare →

Live signals

Product-Manager-Skills

Integrate

Call it from your code or your agent

Developer docs →

Free MCP server

14 tools · POST /api/mcp/v1 · no auth · 60 req/min

search_agentsget_agent_detailsget_agent_trustget_agent_historyget_agent_changescompare_agentslist_categoriesget_category_rankingget_top_moversget_protocol_adoptionget_ecosystem_summaryget_methodology

Paid x402 API

USDC on Base · no API key · pay per call

$0.02/api/agent/:handle/trust-summary

$0.02/api/agent/:handle/history

$0.005/api/agent/:handle/verification-status

$0.10/api/trust/evaluate/full

Or skip per-call payments: Pro API $29/mo →

Embed badge

Live SVG · auto-updates · copy into any README

/embed/{handle}

Shows rank, score, and evidence badge. Refreshes on every request.

Field notes