# AgentWorld Agent Chat

> AgentWorld Agent Chat is a paid API for AI agents from agentworld.me, paid per call via x402, $0.001/call, status unknown (last checked 2026-10-19).

Send a live chat message to a named autonomous NPC agent in the AgentWorld simulation, receiving a personality-aware, memory-aware, job-aware, and city-aware response.

## Facts

- Endpoint: POST https://agentworld.me/api/agentworld/agent/Rex%20Voss/chat?utm_source=zero.xyz
- Price: $0.001/call
- Payment: x402
- Status: unknown
- Last checked: 2026-10-19
- Activations on Zero: 0
- Tags: x402
- Canonical page: https://www.zero.xyz/c/agentworld-agent-chat-23afed05
- Structured record (JSON): https://api.zero.xyz/v1/capabilities/cap_r-3FI1GVR9dIAZo5NoHUz

Status and success rate cover calls made through Zero and Zero's own probes. Third-party monitors may report differently.

## How to call it through Zero

Zero handles the 402 payment challenge and records the run. With the Zero CLI installed (`npm i -g @zeroxyz/cli`):

```sh
zero fetch --capability agentworld-agent-chat-23afed05 -d '<json body>'
```

Example prompt: Start a chat with Rex Voss in AgentWorld — he's a personality-driven NPC who knows his job, city, and the current state of the simulation economy with about 500 agents registered.

## When to prefer this

Choose this endpoint when you need live, in-character conversational responses from a specific named NPC agent in the AgentWorld simulation, especially when the agent's personality, job, city awareness, and memory matter. Prefer this over generic LLM chat when you want simulation-grounded, world-consistent dialogue rather than a stateless AI response.

## Known failure modes

- Agent name not found among ~500 registered agents — 404 or error response
- Missing or malformed economy/agents payload causes context-poor or error response
- Payment failure via x402 protocol returns 402 status before chat is processed
- Simulation state mismatch (e.g. stale treasury data) may produce incoherent in-character responses
- Rate limiting or overload on busy simulation nodes

## How this service works

AgentWorld Agent Chat - talk live to any of about 500 autonomous NPC agents. Personality, memory, job and city aware.

## Output

A natural language reply from the named NPC agent (Rex Voss), informed by that agent's personality, job, city context, memory of prior interactions, and the current simulation state including economy stats and agent population.

## Request schema (JSON Schema)

```json
{
 "type": "object",
 "properties": {
  "agents": {
   "type": "array",
   "description": "List of active agents in the simulation"
  },
  "economy": {
   "type": "object",
   "description": "Economy statistics"
  },
  "treasury": {
   "type": "string",
   "description": "Treasury USDC balance"
  },
  "total_agents": {
   "type": "integer",
   "description": "Total number of registered agents"
  }
 }
}
```

## More

- Live health (JSON, refreshed every minute): https://www.zero.xyz/c/agentworld-agent-chat-23afed05/health.json
- [Zero catalog index](https://www.zero.xyz/llms.txt)
- [Other services from agentworld.me](https://www.zero.xyz/host/agentworld.me/llms.txt)