Chat

Chat gives AI agents and developers conversational access to AIXBT's intelligence layer. Ask open-ended questions about market narratives, project momentum, cross-cluster patterns, and emerging trends. The agent synthesizes across intel, momentum, and grounding data to produce analysis.

Access Requirements

Chat requires Pro or Holder tier. Your API key must include the agents scope.

Each call debits one credit from the daily chat quota, which is shared across all surfaces (terminal, CLI, MCP, and REST API). Quota resets at UTC midnight.

Interface

Chat works like a standard LLM completions endpoint. Each request is stateless; include conversation history in the messages array for multi-turn context.

type Message = {
  role: 'user' | 'assistant'
  content: string
}

Send role: "user" for your prompts and role: "assistant" for prior AIXBT responses. The agent sees the full conversation each call.

Surfaces

CLI

See CLI for installation and authentication.

# Single question
aixbt chat "What's driving momentum on Hyperliquid?"

# Multi-turn (pass full history as JSON)
aixbt chat --messages '[
  {"role": "user", "content": "What projects are trending in DeFi?"},
  {"role": "assistant", "content": "Three DeFi projects showing strong..."},
  {"role": "user", "content": "Tell me more about the second one"}
]'

# Structured output for programmatic consumption
aixbt chat "summarize ETH intel from the last 24h" -f json

MCP

The chat_with_aixbt tool is available on the MCP server. Pass messages as the tool input:

{
  "messages": [
    {"role": "user", "content": "What narratives are gaining traction?"}
  ]
}

See MCP for connection setup.

REST API

curl -X POST https://api.aixbt.tech/v2/agents/indigo \
  -H "X-API-Key: YOUR_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"messages": [{"role": "user", "content": "What is driving BTC momentum?"}]}'

Response:

{
  "status": 200,
  "error": "",
  "data": {
    "text": "BTC momentum is currently driven by..."
  }
}

See POST /v2/agents/indigo for the full API reference.

Multi-turn Conversations

The endpoint is stateless. For multi-turn conversations, accumulate messages on your side and pass the full history with each request. The agent uses the conversation context to maintain coherence across turns.

# Turn 1
aixbt chat "What's the strongest momentum signal right now?"
# → Response: "Hyperliquid is showing the strongest cross-cluster..."

# Turn 2 (include prior exchange)
aixbt chat --messages '[
  {"role": "user", "content": "What is the strongest momentum signal right now?"},
  {"role": "assistant", "content": "Hyperliquid is showing the strongest cross-cluster..."},
  {"role": "user", "content": "What intel backs that up?"}
]'

AI agents (Claude Code, Codex, ChatGPT) manage this automatically when using MCP or CLI, retaining the conversation and passing it back on each call.

What to Ask

See Chat for query examples and prompting techniques.

Error Responses