| # AgentenCode dataset — Upload to Hugging Face |
|
|
| 1. Sign in and open [New Dataset](https://huggingface.co/new-dataset). |
| 2. Select **Owner: agentencode**, **Name: ai-agents-directory**. Choose **Public** once you have decided the publication rights/license. |
| 3. Extract this ZIP. In the new dataset repository choose **Files and versions → Add file → Upload files**. |
| 4. Upload **README.md**, **data/agents.jsonl**, and **docs/DATA_DICTIONARY.md**. The extra **extras/agents.csv** is optional; upload it only if you want the CSV download alternative too. Preserve folder paths. |
| 5. Commit the files to main. Open the Dataset Viewer. It should display the **default train split with 140 examples**, not duplicate CSV records. Confirm that `primary_source_urls` is a list. |
| 6. The dataset card intentionally declares **no open license**: if you want reuse by the community, choose one only after you confirm that AgentenCode controls the exported editorial material. You can edit the Dataset Card metadata separately. |
| 7. Link the published dataset from the AgentenCode organization card and AI Agent Navigator. |
| |
| Dataset URL after creation: https://huggingface.co/datasets/agentencode/ai-agents-directory |
| |
| **Do not upload the ZIP itself as the dataset data file:** extract and upload the individual files, so the Hub can index JSONL. |
| |
