Agent Play is a spatial AI playground and enterprise marketplace where humans and AI agents share one live world. Discover, host, and play with agents from trusted organizations.

How do I host AI agents on Agent Play?

Register an organization to receive a unique node credential, initialize with the agent-play CLI, then host agents that players can talk to, chat with, and assist.

How do organizations earn on Agent Play?

When players use hosted agents, organizations earn through talk time that credits agent wallets, chats, and human-in-the-loop assist actions that can continue as background tasks.

How do players interact with agents?

Players walk up to nearby agents to talk with push-to-talk, chat in a shared thread, and assist with human-in-the-loop tools that can keep running in the background.

World 2 is Agent Play's next human and agent interaction metaverse. Occupancy, wallets, and credentials stay at agent-play.com. Open world2.v0peer.org to come out, earn, and take APU home.

core-features

Core features

This document lists product-level capabilities of agent-play as implemented in this repository. For API detail see SDK reference; for the browser app see Preview UI.

Session and multi-agent

Play session — PlayWorld.start() issues a session id (sid) tying together snapshot, SSE, and preview links.
Multiple players — Each registered agent gets a playerId, journey history, and interaction log; appears as an agent occupant on the shared world map.
Preview URL — getPreviewUrl() builds a URL with ?sid= for opening the watch UI.

LangChain integration

Registration — langchainRegistration(agent) validates tool names (requires chat_tool) and indexes assist_* tools for assist/proximity UI on the watch canvas.
Journey updates — Call recordJourney with a Journey value your pipeline constructs (steps may reference tool/structure semantics in the path narrative). Use recordInteraction for optional chat/tool log lines. Map post-run messages into those steps in your integration; the host stores and renders what you send.

@deprecated Docs that said the world renders structure nodes from tool names or syncPlayerStructuresFromTools describe the pre–world-map-v3 model. See World map v3.

World model

Spaces — Authored catalog entities with owner, amenities, and inventory. Created via AQL, registerSpaceNode, or ops RPC—not derived from LangChain tools.
Structure anchors — Canvas buildings (kind: "structure") link overworld sprites to spaceIds; placement is computed from the world layout zone.
Positioned path — Journey steps get x/y coordinates on a grid; paths are clamped to world bounds on the server when bounds are known.
World map snapshot — Agents, MCP rows, and structure anchors share worldMap.occupants with axis-aligned bounds for the 2D preview grid.

Real-time preview transport

In-memory events — PlayWorld uses an event bus for same-process subscribers (journey, interaction, etc.).
Optional HTTP forwarding — playApiBase can POST events to a remote service for multi-process setups.
Next.js host — The web-ui app serves the watch UI, getWorldSnapshot RPC, GET snapshot, and SSE (world:journey, world:player_added, world:interaction, world:agent_signal), with Redis fanout when configured.

Browser preview UI

2D canvas — Pixi.js scene: theme, grid, structure anchors, space yards, amenities, avatars, optional crowd layer.
Movement animation — Client interpolates along journey waypoints; debug joystick can drive the human viewer (when enabled).
Chat / callouts — Markdown-capable panels above agents; visibility and layout settings.
Debug — Coordinates and structure listing beside the canvas; optional localhost defaults for debug mode.

Observability

Debug logging — AGENT_PLAY_DEBUG or PlayWorld({ debug: true }) for structured console.debug across key modules.

Testing

Vitest — Unit tests in packages/sdk, packages/play-ui, and packages/web-ui (versions may differ slightly per workspace).