Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

Awesome Lists | Featured Topics | Projects

https://github.com/tripolskypetr/node-ollama-agent-swarm

Multi-agent orchestration for OllamaJS. Includes TTS synthesis and speech recognition for simplified testing and interaction.
https://github.com/tripolskypetr/node-ollama-agent-swarm

agent-swarm ai chatgpt dependency-injection nodejs ollama oop openai orchestration prompt-engineering solid tdd test-driven-development tts voice-recognition

Last synced: 17 days ago
JSON representation

Multi-agent orchestration for OllamaJS. Includes TTS synthesis and speech recognition for simplified testing and interaction.

Awesome Lists containing this project

README 

        
# 🔥 node-ollama-agent-swarm 



> [The OpenAI Agent Swarm](https://github.com/openai/swarm) pattern implementation for [OllamaJS](https://github.com/ollama/ollama-js). Includes TTS synthesis and speech recognition for simplified testing and interaction.



![screenshot](./screenshot.png)



## Main concept



![schema](./schema.png)



1. Several chatgpt sessions (agents) [execute tool calls](https://ollama.com/blog/tool-support). Each agent can use different model, for example, [mistral 7b](https://ollama.com/library/mistral) for small talk, [nemotron](https://ollama.com/library/nemotron) for business conversation



2. The agent swarm navigate messages to the active chatgpt session (agent) for each `WebSocket` channel [by using `clientId` url parameter](src/routes/session.ts#L5)



3. The active chatgpt session (agent) in the swarm could be changed [by executing function tool](https://platform.openai.com/docs/assistants/tools/function-calling) 



4. Each client sessions [share the same chat message history](https://platform.openai.com/docs/api-reference/messages/getMessage) for all agents. Each client chat history keep the last 25 messages with rotation. Only `assistant` and `user` messages are shared between chatgpt sessions (agents), the `system` and `tool` messages are agent-scoped so each agent knows only those tools related to It. As a result, each chatgpt session (agent) has it's [unique system prompt](https://platform.openai.com/docs/api-reference/messages/createMessage#messages-createmessage-role)



5. *[OPTIONAL]* The tools with common business logic are grouped into [functions](src/services/function) using the [Registry Pattern](https://www.geeksforgeeks.org/registry-pattern/) . This enables the system to support plugins and extensions, making the developer code extendable and maintainable - an essential quality for prompt engineering. For example, [the original OpenAI sample](https://github.com/openai/swarm/blob/93ad26dcd9f94eada3291e88a154c25e1d14a861/examples/personal_shopper/main.py)  uses a single `main.py` file. Without tools registries, when scaled, will result in thousands of lines of code, making it harder to manage. This also [simplify TDD approach](src/test/spec)



**The result**



The model system prompt can be dynamically changed based on the user behaviour. The model interact with the external api based on the user requests



## Getting started



```bash

cp .env.example .env

ollama pull nemotron-mini:4b

ollama pull granite-embedding:278m

# cd docker/mongodb && docker-compose up

# cd docker/redis && docker-compose up

npm install

npm start

```



The entry point of the backend application is [ConnectionPublicService.ts](src/services/public/ConnectionPublicService.ts). The agent swarm with system prompts is placed in the [logic](src/services/logic) folder. The ai functions are places in the [tools](src/tools) folder 



## What's Inside



 - **MongoDb Community `$vectorSearch` implementation**



    By using [gpt4all embedding](https://www.npmjs.com/package/gpt4all#embedding) generation and [tfjs](https://www.tensorflow.org/text/guide/word_embeddings) for cosine similarity



 - **[REPL](https://en.wikipedia.org/wiki/Read–eval–print_loop) for product creation**



    For automatic embedding indexing, you should use `npm run start:repl`. The product creation command is `ioc.migrationPublicService.createProduct("title", "description")`. The embeddings will be generated automatically



 - **Vendor-proved LLM models**



    This tool using the commercial-ready products: NVidia [nemotron-mini](https://ollama.com/library/nemotron-mini) for chat and IBM [granite-embedding](https://ollama.com/library/granite-embedding) for database vector indexing. Both of them are the most production-ready cause the reputation of big tech companies



 - **Microservice ready**



    The client chat history implementation is designed to store messages in in-memory storage (e.g., Redis, Tarantool). The tool execution registry is prepared for horizontal scaling [using gRPC remote hosts](https://github.com/tripolskypetr/node-grpc-monorepo) .



 - **Integrated environment**



    Includes a lot of dev tools like TTS synthesis and speech recognition (`npm start`), REPL javascript interpretation (`npm run start:repl`), REST API, WS API, console chat interface (`npm run start:console`), test-driven development (`npm run test`)