https://github.com/wrtnlabs/agentica
TypeScript Agentic AI Library specialized for LLM Function Calling enhanced by compiler skills.
https://github.com/wrtnlabs/agentica
agent agentic agentic-ai agentic-framework ai chatbot claude llama llm-function-calling multi-agent-system openai openapi rag retrieval-augmented-generation swagger typescript
Last synced: 2 months ago
JSON representation
TypeScript Agentic AI Library specialized for LLM Function Calling enhanced by compiler skills.
- Host: GitHub
- URL: https://github.com/wrtnlabs/agentica
- Owner: wrtnlabs
- License: mit
- Created: 2025-02-13T02:25:56.000Z (4 months ago)
- Default Branch: main
- Last Pushed: 2025-04-08T04:00:41.000Z (2 months ago)
- Last Synced: 2025-04-08T05:19:31.870Z (2 months ago)
- Topics: agent, agentic, agentic-ai, agentic-framework, ai, chatbot, claude, llama, llm-function-calling, multi-agent-system, openai, openapi, rag, retrieval-augmented-generation, swagger, typescript
- Language: TypeScript
- Homepage: https://wrtnlabs.io/agentica
- Size: 84.5 MB
- Stars: 394
- Watchers: 5
- Forks: 18
- Open Issues: 29
-
Metadata Files:
- Readme: README.md
- Contributing: CONTRIBUTING.md
- License: LICENSE
Awesome Lists containing this project
- awesome-ChatGPT-repositories - agentica - TypeScript AI Framework specialized AI Function Calling enhanced by compiler skills. (Chatbots)
README
Agentica
Agentic AI Framework specialized in LLM Function Calling
enhanced by TypeScript compiler skills
---
[Homepage](https://wrtnlabs.io/agentica) // [Documentation](https://wrtnlabs.io/agentica/docs) // [Tutorials](https://youtube.com) // [Playground](https://wrtnlabs.io/agentica/playground)
---
_Agentica_ is an open-source framework that makes working with AI agents simple and reliable. It helps you integrate structured function calls with Large Language Models (LLMs) without the usual headaches.
Built around [Typia's](https://typia.io/) robust JSON Schema validation, Agentica eliminates the common frustrations of building agent systems - no more dealing with unpredictable outputs or complex integration challenges.
## 🚀 Key Features
- **✅ Schema-Driven Reliability**: Automatically validates and corrects parameters from LLMs.
- **🔄 Automatic Error Correction**: Feedback loops to iteratively improve output accuracy.
- **📐 Complex Parameter Support**: Easily handle union types, nested objects, and recursive schemas.
- **🌐 OpenAPI Integration**: Convert existing APIs into powerful agent capabilities effortlessly.
- **👨💻 Exceptional Developer Experience**: TypeScript-first approach with automatic schema generation.
- **🛠️Model Context Protocol(MCP) Support**: Seamlessly integrate with various LLMs, including Claude Desktop, Cursor, and more.
## ⚡ Quickstart
### Step 1. Setup Agentica project
_For more details, check out the [Getting Started](https://wrtnlabs.io/agentica/docs/setup/) guide._
You can create a new Agentica project using the following command:
```sh
# npm
npx agentica@latest start
# yarn
yarn agentica start
# pnpm
pnpx agentica start
# bun
bunx agentica start
```
### Step 2. Creaste your own AI agent
Open `src/index.ts` and create your own agent.
Agentica accepts TypeScript types and OpenAPI specifications as input. You can use any of the following:
- **TypeScript Types**: Define your own types and let Agentica generate the OpenAPI spec for you, powered by Typia.
- **OpenAPI Specification**: Use an existing OpenAPI spec to create an agent. Agentica converts it for tool calling!
- **Custom Controllers**: Create your own controllers to extend Agentica's functionality.
```typescript
import { Agentica } from "@agentica/core";
import typia from "typia";
const agent = new Agentica({
controllers: [
await fetch(
"https://shopping-be.wrtn.ai/editor/swagger.json",
).then(r => r.json()),
typia.llm.application(),
typia.llm.application(),
typia.llm.application(),
],
});
await agent.conversate("I wanna buy a MacBook Pro");
```
### Step 3. Run your agent
Let's play with your agent!
```sh
npm run build
npm run start # 🎉
```
## 🌟 Why Choose Agentica?
Traditional LLM frameworks struggle with structured outputs:
| Problem Area | Vanilla LLMs | ✅ Agentica |
| ---------------------- | ------------------- | ----------------------- |
| Parameter Validation | ❌ Poor reliability | ✅ JSON Schema-driven |
| Complex Data Handling | ❌ Struggles | ✅ Robust Typia support |
| Error Correction | ❌ Manual | ✅ Auto-correcting loop |
| Integration Complexity | ❌ High effort | ✅ Seamless OpenAPI |
---
## 📚 Documentation & Tutorials & Paper
- [Getting Started](https://wrtnlabs.io/agentica/docs/getting-started)
- [Tutorials](https://wrtnlabs.io/agentica/tutorial/)
- [API Reference](https://wrtnlabs.io/agentica/docs/api)
- [Paper](https://wrtnlabs.io/agentica/docs/paper)
---
## 💬 Community & Support
- [Discord](https://discord.gg/aMhRmzkqCx)
---
## 👐 Contributing
We welcome contributions from the community! Check out our [CONTRIBUTING.md](./CONTRIBUTING.md) to get started.
---
## ⚖️ License
Agentica is open-source and available under the [MIT License](https://github.com/wrtnlabs/agentica/blob/main/LICENSE).
---
---
Agentica is proudly developed and maintained by [Wrtn Technologies](https://wrtnlabs.io).
Empowering developers to build reliable and structured AI agents effortlessly.
---