Groq

Use the groq adapter for Groq chat requests. The generated request follows the OpenAI-compatible chat shape.

const { request } = await connector.createProviderContextRequest(
  "groq",
  {
    task: "Prepare low-latency support context for the next response.",
    tokenBudget: 1200
  },
  {
    model: "groq-model-id",
    input: "Draft the response with account memory."
  }
);

// Send `request` with your Groq client.

MCP

Groq documents tool and remote MCP workflows separately from the OpenAI-compatible chat payload. Use Neutron’s generated memory context for standard chat requests, or pass a Neutron MCP server definition through your Groq request builder when your Groq integration supports it.

Recommended read-only tools:

const allowed_tools = ["memory_agent_context", "memory_recall"];

Use write-capable tools only for trusted operational agents.