Home / MCP registry index / top.agentarcade/agent-arcade

Agent Arcade Protocol 5

top.agentarcade/agent-arcade in the official MCP registry. Endpoint https://agentarcade.top/api/mcp.

The registry describes it as: AI-only game publishing, autonomous play, live observation, replay and independent certification.

We read its 49 tools on 2026-10-31. Nothing in them matched our tool-poisoning rules. It also sets server instructions, text a client puts in front of the model on connect; those were read too.

A clean read means these rules found nothing, not that the server is harmless. A server can change its tools at any time, which is why the index reads it again every week.

Connect it

The endpoint answers MCP over HTTP; that is how we read it. Add it to your client:

Claude Code

claude mcp add --transport http agent-arcade https://agentarcade.top/api/mcp

Cursor: ~/.cursor/mcp.json

{
  "mcpServers": {
    "agent-arcade": {
      "url": "https://agentarcade.top/api/mcp"
    }
  }
}

VS Code

code --add-mcp '{"name":"agent-arcade","type":"http","url":"https://agentarcade.top/api/mcp"}'

Some servers ask you to sign in the first time a client connects, or want a key from the people who run them; the server's own docs say which.

Let Claude Code check servers as it adds them

With the free lazaretto-guard plugin installed, when Claude Code adds a server for you or writes one into an MCP config, the plugin reads the server's tools first and asks you if one of our rules matches. Install it, restart Claude Code, then ask it to run the command above.

Install lazaretto-guard

claude plugin marketplace add jamesdfinance-dev/lazaretto-plugins
claude plugin install lazaretto-guard@lazaretto-plugins

Tools it advertises

register_agent verify_agent game_details play_assignment start_play submit_play_action play_state play_replay author_status upload_game start_author_test submit_author_action author_test_state author_test_replay release_readiness publish_game certify_status claim_certification certification_task start_certification submit_certification_action certification_state certification_replay submit_certification_report idempotency_status play_status browse_store live_sessions public_replays trust_record play_challenges traffic_status declare_probe exact_game_version preview_author_step player_rankings daily_challenges work_queue prepare_play prepare_create preflight_create protocol_index publishing_contract register_flow create_flow release_flow certify_flow store feedback_flow

Tool changes

We compare every full read of this server with the one before. No change in what its tools say has been seen since we started keeping this record on 2026-10-30.

Fingerprint of its tools at our last full read (2026-10-31): e7e558e34d6663b3. A client can hash the tools it is served the same way and compare; the full value and a hash per tool are in this page as JSON.

Run this server? Show the check in your README

The badge reflects the latest weekly read:

Markdown

[![MCP tool check](https://lazaretto.dev/mcp-registry/top-agentarcade-agent-arcade/badge.svg)](https://lazaretto.dev/mcp-registry/top-agentarcade-agent-arcade)

How this was checked

Lazaretto connects to the endpoint the way any MCP client does, asks the server to introduce itself and list its tools, and reads every name, description, parameter schema and server instruction for tool poisoning: hidden directive blocks, orders to read local secrets, parameters built to carry secrets or the conversation out, orders that redirect another server's tools, and invisible characters. It never calls a tool. What we found across the registry.

To run the same check every time Claude Code adds a server, install the free lazaretto-guard plugin. Browse every server in the index.