Orca is an AI agent for deep LinkedIn profile analysis, targeting sales, recruiting, investing, and job seekers. It autonomously scrapes and reasons over LinkedIn data—posts, comments, reactions, network interactions—to extract structured insights like pain points, values, and evolving interests, offering a richer understanding than standard profile views.

How It Works

The agent takes a LinkedIn profile URL and desired insights, scraping core data (profile, posts, comments, reactions, engagement). It then uses a reasoning engine powered by LangChain and configurable LLM providers (e.g., OpenAI) to analyze this data. Orca autonomously triggers further scraping tools if specific insights require additional information, streaming results progressively to the UI.

Quick Start & Requirements

• Installation: Clone (git clone https://github.com/dimimikadze/orca.git), cd orca, pnpm install, pnpm dev. Accessible at http://localhost:3000.
• Prerequisites: Node.js v20+, pnpm. Requires a "Fresh LinkedIn Profile Data" API key and an LLM provider API key (OpenAI default).
• Configuration: Set environment variables in .env.local (RAPIDAPI_KEY, OPENAI_API_KEY). Optional Supabase credentials enable email/password authentication.

Highlighted Details

• Autonomous data acquisition and AI-driven reasoning for deep profile analysis.
• Flexible LLM integration via LangChain, supporting multiple providers.
• Progressive data streaming to the UI as insights are generated.
• Comprehensive test suite for scrapers and agent, with live data or fixture options.

Maintenance & Community

Contribution guidelines are in CONTRIBUTING.md. CI status is via GitHub Actions. No specific community channels or sponsorship details are provided.

Licensing & Compatibility

Distributed under the MIT License, generally permissive for commercial use, modification, and distribution with attribution.

Limitations & Caveats

Relies on external API keys for LinkedIn data and LLMs, potentially incurring costs. Optional Supabase integration is needed for authenticated access; otherwise, the app runs openly.