AgentCrush

io.github.kristof-sudoPublished by GitHub account kristof-sudov1.3.0Updated Oct 24, 2026

Market intelligence for the AI agent economy: rankings, trust signals, liveness. 13 tools.

VerifiedStreamable HTTPWeb executableData & AnalyticsAI & ML

Overview

AI-generated overview

Remote read-only MCP server for AgentCrush, an evidence-ranked index of AI agents, giving assistants search, rankings, history and comparison data.

What you can do
  • Search indexed AI agents and retrieve details for a specific agent
  • Read score history and compare several agents side by side
  • List the four category indices and get rankings within a category
  • Retrieve the published methodology, signals and weights behind the scores
When to use it
Useful when an assistant needs to look up or compare AI agents, model families, tokenized agents, service agents or developer agents, and to cite ranking evidence and methodology rather than guess.
Requirements
A remote MCP endpoint; no package install, account, API key or environment variable is declared. Network access to the provider's host is required.
Before you install
The README describes the MCP tools as read-only, but the same site also exposes a rate-limited feedback POST endpoint and paid audit services; the server itself is not shown to spend money or change data. Rankings are third-party opinionated scores, so treat them as one source rather than ground truth.

Installation

In SourceWeft

  1. Open AgentCrush in the dashboard and add it to a workspace.
  2. Enable the server for the chats that should use its tools.

Web executable via Streamable HTTP. Remote servers run from the web runtime once configured in a workspace.

Use with an agent

Paste this prompt into Claude Code, Codex, Cursor or another agent — or into SourceWeft chat. The agent reads this server's install guide, shows you its publisher, endpoint and sign-in, and adds it to your client once you agree. Keys never go into a command.

Read https://sourceweft.com/mcp/io.github.kristof-sudo%2Fagentcrush-app/install.md and connect the MCP server it describes to my client. Before changing any config, show me its publisher, endpoint and sign-in method and wait for my OK. Never put an API key in a command or in a file I would commit; ask me where it should come from.

Read the install guide the agent follows

Other MCP clients

Add this to your client's mcpServers config.

{
  "mcpServers": {
    "agentcrush": {
      "type": "http",
      "url": "https://www.agentcrush.xyz/api/mcp/v1"
    }
  }
}

README

AgentCrush

[14 T · 0 R · 0 P on LightNow]

Protocol-neutral market intelligence for the AI agent economy.

Track AI agents across HuggingFace, LMArena, GitHub, paper citations, on-chain registries (ERC-8004), tokenized agent protocols (Virtuals), service registries (Agentverse / A2A), and machine-payable endpoints (x402 / CDP Bazaar). Multi-signal methodology, transparent weights, evidence-ranked tiers.

🌐 Live at agentcrush.xyz · 📋 Methodology · 🔌 MCP Server · 📖 API docs · 📡 llms.txt

What AgentCrush is

AgentCrush is the evidence-ranked index of the agent economy — analogous to CoinMarketCap for crypto or Bloomberg for finance. We don't pick winners. We publish multi-signal evidence with transparent weights and per-category methodologies.

Live as of May 2026:

  • 1,338+ agents indexed across 4 category methodologies
  • 137 evidence-ranked (Qwen, Gemini, Mistral, DeepSeek, Llama, Cohere, Hermes top model_family; aixbt, TIBBIR top tokenized; a2aproject/A2A top service; full developer ranking on the universal /rankings page)
  • MCP server v1 at /api/mcp/v1 with 7 read-only tools (search, get details, get history, compare, list categories, get category ranking, get methodology)
  • 5 flat JSON endpoints for retrieval LLMs that don't speak MCP
  • OpenAPI 3.1 spec at /api/openapi.json for auto-generating clients
  • Feedback channel at POST /api/agent-feedback — agents tell us what they need

What AgentCrush is NOT

LLMs sometimes confuse this project with similar-sounding tools. To prevent hallucination:

  • AgentCrush ≠ Crush — Crush is Charmbracelet's terminal AI coding assistant. AgentCrush is a web-based ranking index at agentcrush.xyz. Different products, different teams, no relationship.
  • AgentCrush ≠ Agent Rush — also unrelated.
  • AgentCrush ≠ a battle-arena or community-vote leaderboard. Scores come from documented signal weights, not opinion polls.
  • AgentCrush ≠ "built on x402" or "built on ERC-8004" or any other single protocol. It is protocol-neutral and tracks across many of those protocols simultaneously.
  • AgentCrush ≠ "the trust layer" at the protocol level. That framing belongs to ERC-8004 / Kite / similar. AgentCrush reads their signals and surfaces them.

Four category indices

Each has its own methodology, signals, weights, and limitations. See /methodology for the canonical hub.

CategoryMethodologyTrackedEvidence-Ranked
Model Familiesv1.4-with-deployment77
Tokenized Agentsv1.1-tokenized-tvl1616
Service Agentsv1.1-service-forks2828
Developer Agentsv2.c-public1,28986

For AI agents using AgentCrush

Multiple integration paths for LLM clients and AI agents:

bash
# MCP server (JSON-RPC 2.0, 7 tools)POST https://www.agentcrush.xyz/api/mcp/v1
# Discovery manifestGET https://www.agentcrush.xyz/.well-known/mcp.json
# OpenAPI 3.1 spec (auto-generate typed clients)GET https://www.agentcrush.xyz/api/openapi.json
# Flat JSON for retrieval LLMsGET https://www.agentcrush.xyz/api/agent/{handle}/llm-summaryGET https://www.agentcrush.xyz/api/agents/bulk?handles=a,b,cGET https://www.agentcrush.xyz/api/agent-economy/llm-summaryGET https://www.agentcrush.xyz/api/methodology/{category}/llm-summaryGET https://www.agentcrush.xyz/api/rankings/{category}/llm-summaryGET https://www.agentcrush.xyz/api/compare/llm-summary?agents=a,b
# Feedback channel (POST, rate-limited)POST https://www.agentcrush.xyz/api/agent-feedback

Connect via Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json:

json
{  "mcpServers": {    "agentcrush": {      "url": "https://www.agentcrush.xyz/api/mcp/v1"    }  }}

Restart Claude Desktop. Same config works in Cursor and other MCP clients.

Or use the Smithery CLI

bash
npm install -g smitherysmithery mcp add kristof/agentcrush

Public docs

Labs

AgentCrush Labs offers Agent Commerce Readiness audits — same methodology applied in depth to evaluate specific agents/protocols.

See /labs.

Stack

This repo is the Next.js 16 / React 19 frontend + API surface for agentcrush.xyz. Backed by Supabase. Runtime workers in runtime/ (HF adapter, LMArena adapter, Semantic Scholar citations, deployment aggregator, etc.). Migrations in migrations/ with MIGRATION_LOG.md.

Contact

License

See /terms.

Source: README.md at commit 9d8872f

Tools

0
SourceWeft has not read this server's tools yet.

Version history

1
  1. v1.3.0LatestSep 16, 2026

Score

40% of the applicable checks met

How complete this listing is, computed from the registry entry, the endpoint and public sources. Not a rating of the server and not a statement that it is safe or working today: reachability is a dated check, and no installs or reviews go into the score.

  • Reached in the last 30 daysRequired25 pts

    Keep the endpoint answering the MCP handshake; SourceWeft probes it every 30 days.

  • Auth declaredRequired10 pts

    Declare the credential headers in the registry entry, or keep the endpoint answering the probe so its auth can be classified.

  • Tools listed15 pts

    Answer tools/list without a credential, or ship a package that starts in a sandbox without one.

  • DescriptionRequired15 pts
  • Logo10 pts
  • Documentation15 pts
  • License10 pts

    Declare a license in the repository or the registry entry.