Inception ICORE Server

What is Inception ICORE Server?

A modular, extensible Rust-based server providing short-term, long-term, and latent memory services, a chat endpoint backed by a BaseAgent + Sentience DSL, and seamless integration with ChromaDB and LLM services. It is designed as a flexible microservice for building AI systems with layered memory and agent logic.

How to use Inception ICORE Server?

Clone the repository, build with cargo build, set environment variables (CHROMADB_URL, CHROMA_COLLECTION_ID, LLM_URL, ICORE_ENV), then run with docker-compose -f docker-compose.dev.yml up -d --build or run natively. Test the server with curl -i http://localhost:8080/api/ping expecting a "pong" response.

Key features of Inception ICORE Server

• Modular memory layers (short-term, long-term, latent via ChromaDB)
• BaseAgent with remember/recall/context logic
• Sentience DSL for scripting custom response logic
• LLM fallback integration (HTTP-based)
• Docker Compose orchestration with ChromaDB and LLM
• RESTful JSON API (Axum, Tokio async)

Use cases of Inception ICORE Server

• Build chatbots with persistent and ephemeral memory
• Prototype AI agents using Sentience DSL scripting
• Manage and query vector embeddings for similarity search
• Enable fallback to local LLM for text generation
• Create modular AI systems with separate memory layers

FAQ from Inception ICORE Server

Is Inception ICORE Server ready for production?

No, it is a work in progress under active development; interfaces, APIs, and internal structure may change frequently.

What are the runtime dependencies?

Rust toolchain, Docker (for ChromaDB and LLM), a ChromaDB service, and an HTTP-based LLM server (e.g., llama.cpp).

Where is data stored?

Short-term memory is in-memory (volatile), long-term in SQLite file (memory.db), latent memory in ChromaDB.

What transport protocol does the API use?

HTTP with JSON request/response, built on the Axum framework.

Are there any known limitations in the current prototype?

Latent memory embeddings are dummy zero-vectors; real LLM encoder integration is not yet implemented.