Mira Core Library

The core library implements the multi-agent system for company research. It is framework-agnostic and provides agents, services, and orchestration logic that can be consumed by other packages such as the demo frontend.

How It Works

• Agents (src/agents/) Specialized modules for tasks like discovery, internal page crawling, LinkedIn lookup, Google Search, and company analysis.
• Orchestrator (src/orchestrator/) Coordinates agents, merges their outputs, manages sources, and tracks progress events.
• Services (src/services/) Wrappers for ScrapingBee (website crawling and Google Search), LinkedIn company scraper, and other helpers.
• Types (src/types/) Shared TypeScript definitions.
• Constants (src/constants/) Centralized configuration of data points and their mapping to internal pages, LinkedIn, and Google Search.
• Entry Point (src/index.ts) Provides the main runEnrichment function to execute the full research workflow.
• Types (src/types.ts) Exposes the library's public type definitions.

Requirements

• Node.js v18+ (LTS recommended)
• Environment variables required for running tests, stored in .env.test:

OPENAI_API_KEY=sk-xxxx
SCRAPING_BEE_API_KEY=xxxx

Testing

This package uses Jest with targeted scripts per agent/service.

Run a specific agent test, for example:

npm run test:discovery-agent

See package.json for other available test scripts for services and agents.

Build

From the root of the repository, run:

npm run build:mira

This compiles the TypeScript source into the dist/ directory.