Evo.ninja is a versatile, generalist AI agent designed to adapt its behavior in real-time to specific tasks by adopting specialized "personas." It's targeted at users who need an adaptable AI assistant for tasks ranging from data analysis and research to software development. The core benefit is its dynamic persona selection, allowing it to optimize its approach for different problem domains.

How It Works

Evo.ninja operates on a four-step execution loop. It first predicts the best next step, then selects an agent persona (e.g., Synthesizer for text, CSV Analyst for tabular data, Researcher for web, Developer for code) that best fits the predicted step. It contextualizes the chat history to include only relevant messages for the current step and finally evaluates and executes the chosen agent's function to progress towards the user's goal. This iterative, persona-driven approach allows for dynamic task adaptation.

Quick Start & Requirements

• Install: Clone the repository, copy .env.template to .env, and add your OpenAI and SerpApi API keys.
• Prerequisites: Git, Node.js (via nvm), Yarn, Docker Desktop (for UI), OpenAI API Key, SerpApi Key.
• Setup: yarn && yarn build to build. Run via CLI with yarn start [goal]. UI requires yarn db:start and yarn dev in apps/browser.
• Docs: Website

Highlighted Details

• Dynamic agent persona selection for task-specific adaptation.
• Supports personas for text synthesis, CSV analysis, web research, and Python development.
• Execution loop includes prediction, persona selection, context refinement, and action execution.
• Session management with dedicated workspaces and internal logging.

Maintenance & Community

• Community support available via Discord.
• Issue reporting via GitHub Issues.

Licensing & Compatibility

• The repository does not explicitly state a license in the provided README.

Limitations & Caveats

The UI component requires Docker Desktop and a Supabase setup, which can add complexity to local deployment. The project relies heavily on external API keys (OpenAI, SerpApi), incurring costs and external dependencies.