diff --git a/README.md b/README.md
index 9a2a17c..e68c181 100644
--- a/README.md
+++ b/README.md
@@ -186,16 +186,23 @@ python scripts/build_rag_index.py
 
 ---
 
-## Kaggle Notebook
+## Notebook
 
-Explore the dataset, RAG corpus, and a prototype Critic → Retriever → Fixer pipeline on Kaggle:
+The project's experimentation and prototyping notebook is available both on Kaggle and as a local export:
 
-https://www.kaggle.com/code/omchoksi04/codereview
+| Resource | Link |
+|----------|------|
+| **Kaggle Notebook** | [https://www.kaggle.com/code/omchoksi04/codereview](https://www.kaggle.com/code/omchoksi04/codereview) |
+| **Local Export** | [`notebooks/codereview.ipynb`](notebooks/codereview.ipynb) |
+| **Notebook README** | [`notebooks/README.md`](notebooks/README.md) |
 
 The notebook demonstrates:
-- Dataset exploration (28,548 records from CodeXGLUE + OWASP)
-- RAG corpus embedding and similarity search
-- Prototype review pipeline
+- Dataset loading and exploration (28,548 records from CodeXGLUE + OWASP)
+- RAG corpus inspection and semantic search
+- Critic → Retriever → Fixer prototype pipeline
+- Evaluation examples
+
+See [`notebooks/README.md`](notebooks/README.md) for full details.
 
 ---
 
diff --git a/docs/demo.md b/docs/demo.md
index e222dc1..304b43c 100644
--- a/docs/demo.md
+++ b/docs/demo.md
@@ -97,9 +97,11 @@ curl -X POST https://OMCHOKSI108-codereview-agent.hf.space/rag/search \
 
 ## Kaggle Notebook
 
-Explore the dataset and prototype pipeline on Kaggle:
+Explore the dataset, RAG corpus, and prototype Critic → Retriever → Fixer pipeline:
 
-https://www.kaggle.com/code/omchoksi04/codereview
+- **Kaggle**: https://www.kaggle.com/code/omchoksi04/codereview
+- **Local export**: [`notebooks/codereview.ipynb`](../notebooks/codereview.ipynb)
+- **Notebook docs**: [`notebooks/README.md`](../notebooks/README.md)
 
 ## Screenshot Checklist
 
diff --git a/docs/project_status.md b/docs/project_status.md
index 2a731ad..5a5e41e 100644
--- a/docs/project_status.md
+++ b/docs/project_status.md
@@ -5,7 +5,7 @@
 | Component | Status | Notes |
 |---|---|---|
 | Dataset (CodeSecAudit-RAG) | **Done** | 28,548 records, 2,833 RAG chunks, published to HF + Kaggle |
-| Kaggle Notebook | **Done** | Dataset exploration + RAG prototype |
+| Kaggle Notebook | **Done** | [Dataset exploration + RAG prototype](../notebooks/README.md) |
 | Review Engine (core) | **Done** | Critic, fixer, retriever, risk scorer, pipeline |
 | GitHub Action | **Done** | PR triggers, summary + inline comments, max caps |
 | Inline PR Comments | **Done** | 10-comment cap, duplicate detection via fingerprint |
diff --git a/notebooks/README.md b/notebooks/README.md
new file mode 100644
index 0000000..a5a2cff
--- /dev/null
+++ b/notebooks/README.md
@@ -0,0 +1,36 @@
+# Notebooks
+
+## CodeSecAudit AI Dataset + RAG Exploration
+
+| Field | Value |
+|-------|-------|
+| **Title** | CodeSecAudit AI Dataset + RAG Exploration |
+| **Author** | Om Choksi |
+| **Platform** | Kaggle |
+| **Kaggle Link** | [https://www.kaggle.com/code/omchoksi04/codereview](https://www.kaggle.com/code/omchoksi04/codereview) |
+| **Local Export** | [`codereview.ipynb`](codereview.ipynb) |
+
+### What it demonstrates
+
+The notebook is the experimentation phase of the CodeSecAudit AI system. It proves the core dataset + RAG + reviewer logic end-to-end:
+
+1. **Dataset Loading** — Loads the `OMCHOKSI108/CodeSecAudit-RAG` dataset (28,548 records from CodeXGLUE + OWASP Benchmark Python) directly from Hugging Face.
+2. **CodeSecAudit-RAG Exploration** — Explores dataset quality, source distributions (CodeXGLUE vs OWASP), language balance (C vs Python), label distributions (vulnerable vs clean), CWE coverage, severity breakdown, and OWASP categories using interactive Plotly visualizations.
+3. **RAG Corpus Inspection** — Loads and inspects the 2,833 OWASP Cheat Sheet Series chunks that power the retrieval-augmented guidance.
+4. **Semantic Search** — Builds a ChromaDB vector index using `all-MiniLM-L6-v2` embeddings (384-dim) and runs similarity search queries against the OWASP corpus.
+5. **Critic → Retriever → Fixer Prototype** — Runs a lightweight end-to-end pipeline: the Critic scans code for CWE patterns, the Retriever fetches relevant OWASP guidance, and the Fixer generates suggested remediations.
+6. **Evaluation Examples** — Tests the pipeline against known vulnerable and safe code samples with structured output.
+
+### Pipeline Path
+
+```
+Notebook prototype → CLI reviewer → FastAPI API → Streamlit UI → GitHub PR bot → Docker deployment
+```
+
+### File
+
+- `codereview.ipynb` — Exported Kaggle notebook (local copy for reference)
+
+### Note
+
+The live notebook is hosted on Kaggle. The local `.ipynb` export may lag behind the latest Kaggle version. Run or fork the notebook directly on Kaggle for the most up-to-date interactive experience.