Add guide for automating code review with OpenCode and Inference Providers #2023
Conversation
|
The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update. |
There was a problem hiding this comment.
Really great guide! Remember to upload the images to https://huggingface.co/datasets/huggingface/documentation-images + link them in the guide (+1 if you have light and dark mode images but not required).
|
|
||
| Here's a real example from a [fork of the Hugging Face datasets repository](https://github.com/davanstrien/datasets). The issue requests adding `uv` installation support: | ||
|
|
||
|  |
There was a problem hiding this comment.
image needs to be upload to HF dataset https://huggingface.co/datasets/huggingface/documentation-images
|
|
||
| When someone comments `/oc fix this`, OpenCode analyzes the issue, creates a new branch, implements the changes, and opens a pull request: | ||
|
|
||
|  |
|
|
||
| The PR includes all the necessary changes: | ||
|
|
||
|  |
| @@ -0,0 +1,182 @@ | |||
| # Automating Code Review with GitHub Actions | |||
|
|
|||
| [OpenCode](https://opencode.ai) is an AI coding agent that runs in your terminal and can work with open models via Hugging Face Inference Providers. | |||
There was a problem hiding this comment.
I think it'd be cool to showcase how to use opencode locally in your terminal as well. I like the fact that the main topic of the guide is about GitHub Actions but I do think a first "local opencode" example would be great.
| 2. **Selecting a provider** - Choose **Hugging Face** from the list | ||
| 3. **Choosing a model** - Select a model like **GLM-4.5-Air** or **Kimi-K2-Instruct** | ||
| 4. **Creating the workflow** - Generates `.github/workflows/opencode.yml` automatically | ||
|
|
There was a problem hiding this comment.
Maybe here a screenshot of what to except from local opencode in terminal?
cc @hanouticelina @Wauplin
Adds a guide on using OpenCode to set up GitHub actions using IP to run the models under the hood.
Toy PR example here: davanstrien/datasets#2 (note I've killed this token now so it won't work if you try it not but I can add you to a private repo if you want to test!
Still to do: