Documentation

# Agents Overview

Explore this guide to learn how ChatboxAI works and how to integrate it into your workflow.

Back to overview

Agents Overview

Build autonomous AI agents that can plan, use tools, maintain memory, and complete multi-step tasks.

What is an Agent?

An agent is an AI system that:

Receives a high-level goal from a user
Plans a sequence of steps to accomplish it
Uses tools (functions, APIs, search) to gather information
Maintains state across multiple turns
Delivers a final result

Basic Agent Pattern

JAVASCRIPT

async function runAgent(goal, tools) {
  const messages = [
    { role: "system", content: "You are an autonomous agent. Use tools to accomplish the user's goal." },
    { role: "user", content: goal }
  ];

  while (true) {
    const response = await client.chat.completions.create({
      model: "chatboxai/gpt-oss-120b",
      messages,
      tools,
      tool_choice: "auto",
    });

    const choice = response.choices[0];
    messages.push(choice.message);

    if (choice.finish_reason === "stop") {
      return choice.message.content; // Agent is done
    }

    if (choice.finish_reason === "tool_calls") {
      for (const toolCall of choice.message.tool_calls) {
        const result = await executeTool(toolCall);
        messages.push({
          role: "tool",
          tool_call_id: toolCall.id,
          content: JSON.stringify(result),
        });
      }
    }
  }
}

Agent Tools

Agents become powerful when given tools:

JAVASCRIPT

const agentTools = [
  {
    type: "function",
    function: {
      name: "web_search",
      description: "Search the web for current information",
      parameters: {
        type: "object",
        properties: {
          query: { type: "string", description: "Search query" }
        },
        required: ["query"]
      }
    }
  },
  {
    type: "function",
    function: {
      name: "read_file",
      description: "Read contents of a file",
      parameters: {
        type: "object",
        properties: {
          path: { type: "string" }
        },
        required: ["path"]
      }
    }
  }
];

Memory Types

Memory Type	Storage	Scope	Use case
In-context	Messages array	Single session	Conversation history
External	Database	Cross-session	User preferences, facts
Semantic	Vector DB	Long-term	Knowledge retrieval

Agent Safety

Always validate tool outputs before passing back to the model
Set a maximum iteration limit to prevent infinite loops
Log all tool calls and model decisions for debugging
Use human-in-the-loop for high-stakes actions

Next Steps

Agent Definitions — Define agent roles and personas
Orchestration — Multi-agent coordination
Workflows — Structured task pipelines