About React + TypeScript + Vite
evoAI is a local Ollama runner that gives LLMs agent-like abilities, including planning tasks and creating tools when needed. It is for users who want local models to build and execute multi-step workflows.
Key features
- 01Runs LLMs through Ollama locally
- 02Plans multi-step tasks from a goal
- 03Generates and registers tools on demand
- 04Executes tools in sequence with error handling
Use cases
- 01Run a multi-step task using a local model
- 02Generate a missing tool during task execution
Configuration
Local server config is the same for every client. Copy it into your client’s MCP config file.
We haven’t collected a package or remote endpoint for this server. Follow the instructions in the project README.
Tools
Local servers expose tools only once running on your machine, and the maintainer has not provided a tool list.
Details
From the project READMEevoAI: LLMs + evolution
evo-ai is a local runner for Ollama that enhances any LLM with agentic capabilities — enabling models not just to use tools, but to create new ones autonomously and invoke them when needed.
💡 Think of it as a LLM which has real-time, self-expanding toolchains built on-demand by the LLM itself.
Features: 🛠️ Tool Creation On the Fly When the LLM encounters a missing tool or function, it can generate and register new code snippets or scripts to accomplish the task.
Modular & Local: No cloud dependencies. Everything runs locally via Ollama, ensuring privacy and full control.
Supports Any LLM: Works with any LLM accessible via Ollama (e.g., LLaMA, Mistral, Code LLMs, etc).
⚠️ Note: The effectiveness of this application heavily depends on the capabilities of the LLM you choose. For best results, use models that excel at coding and analytical reasoning.
📦 Installation
Prerequisites:
git clone https://github.com/pahulgogna/evoAI.git
cd evoAI
npm install
🧪 Usage
Run the agent:
Give it a goal and let the model break it down, create tools, and execute steps autonomously.
🧠 How It Works
Introspective Planning The LLM receives your high-level goal and plans a multi-step solution.
Tool Audit Checks available tools. If something’s missing, it builds it.
Execution Executes tools in sequence, handling input/output passing and error catching.
Self-Extension If new needs arise during execution, the agent can loop back and generate new capabilities on the fly.
🔒 Security Note Since the LLM can generate and execute arbitrary code, it's strongly recommended to run in a sandboxed or containerized environment if you're testing unfamiliar prompts.
🛠️ Project Structure
evoAI/
├── src/
├── electron/
├── ui/
🤝 Contributing Pull requests are welcome! For major changes, please open an issue first to discuss what you’d like to change.
Comments
New to MCP?
From what the protocol is to installing your first server, these answers cover most first questions.
An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.