This `README.md` is designed to reflect the sophisticated local RAG pipeline you've built, highlighting the multi-threaded enrichment and the DSPy-powered "Smart Retrieval" system.

---

# 🐉 DnD Campaign Oracle: Local RAG Assistant

An advanced Retrieval-Augmented Generation (RAG) system designed for Dungeon Masters. This tool ingests markdown-based campaign notes, enriches them with AI-generated metadata, and provides an interactive terminal interface to query your world’s lore using **DSPy** and **Local LLMs**.

## ⚔️ Key Features

* **Parallel Enrichment:** Utilizes a `ThreadPoolExecutor` to process multiple document chunks simultaneously across local LLM slots for high-speed ingestion.
* **Structured Metadata:** Uses **DSPy TypedPredictors** and **Pydantic** to force LLMs to output valid JSON synopses, tags, and entity lists.
* **Deep Context Retrieval:** Unlike standard RAG, this system retrieves relevant chunks and then "peeks" at the full source file to provide the LLM with broader narrative context.
* **Local-First:** Designed to run entirely on your hardware using **LM Studio** and **FAISS**, keeping your campaign secrets private.

---

## 🏗️ Architecture

1. **Ingestion:** Scans `DATA_DIR` for `.md` files.
2. **Chunking:** Splits documents into 800-character segments with overlap.
3. **Enrichment:** A DSPy `IngestionAgent` analyzes each chunk to extract:
* **Synopsis:** A one-sentence summary.
* **Tags:** Plot points, item names, or themes.
* **Entities:** Specific NPCs, Locations, or Factions.


4. **Vector Store:** Chunks and metadata are embedded using `text-embedding-qwen3` and stored in a local **FAISS** index.
5. **Interactive RAG:** A terminal loop that uses **Chain of Thought (CoT)** reasoning to answer queries based on retrieved context.

---

## 🛠️ Setup

### Prerequisites

* **Python 3.10+**
* **LM Studio:** Running a local server at `http://192.168.0.49:1234` (or your specific IP).
* **Models:** * Inference: `qwen3-8b` (or similar).
* Embedding: `text-embedding-qwen3-embedding-8b`.



### Installation

```bash
uv sync

```

---

## 🚀 Usage

### 1. Ingest & Enrich

Run the ingestion script to process your markdown files and build the vector database.

```bash
uv run src/ingest.py

```

### 2. Query the Oracle

Launch the interactive session to ask questions about your campaign.

```bash
uv run src/retrieve.py

```

**Example Query:**

> `📝 Query: Why did the party get free bread at the Golden Grain Inn?`
> `📜 AI RESPONSE: Based on the session notes from 'Session_12.md', the party received free bread because the Rogue successfully intimidated the baker's assistant, and the Cleric later performed a minor miracle (Thaumaturgy) that impressed the owner.`

---

## 📂 File Structure

* `ingest.py`: Handles file loading, multi-threaded enrichment, and FAISS storage.
* `retrieve.py`: The interactive terminal-based retrieval loop.
* `experts/ingestion_agent.py`: Contains the `IngestionAgent` and Pydantic schemas.
* `embedding.py`: Custom wrapper for `LocalLMEmbeddings` with batch processing support.
* `local_faiss_db/`: Directory where the vector index and metadata are persisted.

---

## ⚙️ Configuration

In `ingest_notes.py`, you can tune the processing speed:

* `max_workers=8`: Adjust based on your GPU/CPU capability to handle concurrent LLM requests.
* `chunk_size=800`: Increase for more context per chunk, decrease for more granular searching.

---