Skip to main content

Research Plugin

Research tooling for web-enabled AI research using Gemini models with web search capabilities.

Overview​

The @tokenring-ai/research package provides research tooling for the Token Ring ecosystem. It exposes a research tool that dispatches web-enabled AI model requests (Gemini 2.5 Flash Web Search) to generate research on given topics and prompts.

Key Features​

  • Web search-backed research via Gemini models
  • Strongly-typed parameters and results
  • Token usage logging
  • Cost estimation
  • Integration with AI Model Registry

Core Components​

Tools​

research: Dispatches research request to AI agent

  • Input: { topic: string, prompt: string }
  • Returns: Generated research content with status
  • Uses Gemini 2.5 Flash Web Search model

Result Types:

  • Success: { status: 'completed', topic: string, research: string, message: string }
  • Error: { status: 'error', topic: string, error: string, message: string }

Global Scripting Functions​

When @tokenring-ai/scripting is available:

  • getResearchModel(): Gets the configured research model name 
    /var $model = getResearchModel()
    /echo Using research model: $model

Usage Example​

import { execute as researchExecute } from '@tokenring-ai/research/tools/research';

// Via tool
const result = await agent.executeTool('research', {
  topic: 'Large Language Models',
  prompt: 'Compare safety techniques and cite recent sources'
});

if (result.status === 'completed') {
  console.log(result.research);
}

// Direct API usage
const result = await researchExecute({
  topic: 'Climate Tech Startups',
  prompt: 'Summarize funding trends in 2024 and notable companies'
}, agent);

Configuration Options​

  • Model: Uses Gemini 2.5 Flash Web Search by default
  • Token Limits: Configured via AI model settings
  • Temperature: Controlled by AI configuration

Requirements​

  • Chat service registered in the Registry
  • ModelRegistry service with Gemini model supporting web search
  • Appropriate credentials for Gemini provider

Dependencies​

  • @tokenring-ai/agent@0.1.0: Core agent framework
  • @tokenring-ai/ai-client@0.1.0: AI model integration
  • @tokenring-ai/scripting@0.1.0: Optional, for global functions

Notes​

  • Requires Gemini model with web search capability
  • Token usage data logged when available
  • Status messages appear in interactive sessions
  • Research results include web-sourced information