Update readme.md

README.md

@@ -39,4 +39,57 @@ tags:
 - mcp
 size_categories:
 - 10K<n<100K
----
+---
+
+# n8n Automation Atlas: 36,405 n8n Workflows
+
+A curated collection of **36,405 unique n8n automation workflows**, organized and cleaned for AI training, research, and community use. This dataset combines high-quality synthetic workflows generated via a custom archetype engine with a massive pool of cleaned community templates.
+
+## 🚀 Project Links
+
+- **Main Repository:** [GitHub - Ker102/n8n-workflows-36k](https://github.com/Ker102/n8n-workflows-36k)
+- **Author:** [Ker102 on GitHub](https://github.com/Ker102)
+- **Explorer UI:** A Vue-based web application is included in the main repository to browse these workflows locally.
+
+## 📊 Dataset Statistics
+
+- **Total Unique Workflows:** 36,405
+- **Source Breakdown:**
+    - **Repository & Synthetic:** ~12,500 unique functional scripts.
+    - **Community (Cleaned):** ~23,900 unique community-contributed templates.
+- **Format:** JSON / JSON Lines (.jsonl)
+- **License:** Apache 2.0
+
+## 🧠 Methodology: The Combinatorial Archetype Engine
+
+The majority of the synthetic workflows in this dataset were generated using a **Combinatorial Archetype Engine**. This engine ensures high validity and diversity by using mathematically proven combinations of compatible nodes.
+
+### How it works:
+- **Equivalency Groups:** Nodes are grouped by function (e.g., Databases, Messaging, AI, DevOps).
+- **Archetypes:** Verified "skeletons" of common automation patterns (e.g., "Form to Database & Notify", "AI Summarizer", "RAG Ingestion").
+- **Combinatorial Expansion:** The engine iterates through every valid combination of nodes for each archetype, multiplying the output volume while maintaining structural integrity.
+
+## 🛠️ Data Quality & Cleaning
+
+The community portion of this dataset (sourced from the `n8n_workflows_templates_dataset`) underwent an intensive cleaning process:
+- **JSON Normalization:** Stringified JSON values were parsed into valid nested objects.
+- **Structural Verification:** Every workflow was checked for the presence of `nodes` and `connections`.
+- **Deduplication:** Workflows were deduplicated based on unique structural fingerprints and IDs.
+- **Metadata Extraction:** Node counts and integration types were extracted to allow for easy filtering.
+
+## 📂 Directory Structure (in GitHub)
+
+- `workflows/`: Contains the raw JSON files for each workflow.
+- `web/`: A Vue + Vite application serving as a "Workflow Explorer" UI.
+- `scripts/`: Tools for manifest generation, synthetic creation, and data cleaning.
+
+## 🤝 Community & Support
+
+This project is maintained by **Ker102**. If you find this dataset useful, please consider:
+- Giving the [GitHub repository](https://github.com/Ker102/n8n-workflows-36k) a ⭐.
+- Following [Ker102 on GitHub](https://github.com/Ker102) for more automation and AI-related projects.
+- Contributing new archetypes or reporting issues in the GitHub repository.
+
+---
+
+*Note: These workflows are provided as-is for research and development purposes. Always review workflows before importing them into a production n8n environment.*