RAMA - pull Request Augmentation for Model Awareness

Automated analysis of model and metamodel changes in GitHub Pull Requests.

RAMA is a GitHub Actions-based tool that detects, analyzes, and reports changes in EMF models and metamodels in pull requests. Models are compared at model-level via EMF Compare, and munidiff-based (1, 2) difference reports are automatically included as comments in the analyzed pull request discussion.

Quick start

RAMA is designed to run in the repository whose pull requests it analyzes.

1. Add a rama.json file to the root of the target repository. Start with the configuration below and adapt its model extensions and metamodel paths.
2. Copy rama.example.yml to .github/workflows/rama.yml in that repository.
3. Open or update a pull request that modifies a configured model or metamodel file. The workflow builds RAMA, retrieves the changed files through the GitHub API, and posts its report as a pull request comment.

The workflow template clones this repository from https://github.com/munidiff/rama. If you use a fork of RAMA, replace that URL .github/workflows/rama.yml.

Configuration

RAMA has to be configured in the target repository. The target repository is the repository where the GitHub Action runs and whose pull requests RAMA analyzes. The workflow reads rama.json from the checked-out target branch, so configuration changes made only in a pull request take effect after they are merged.

If the target-branch revision does not contain a rama.json file, or the file is empty or invalid JSON, RAMA uses its packaged default configuration and adds a configuration warning to the pull-request comment. If you want RAMA to analyze extensions or metamodels other than the defaults, add a valid rama.json file at the root of the repository.

The file must use this structure. The following example shows RAMA's default configuration:

{
  "model_extensions": [".model"],
  "metamodels": []
}

• model_extensions: file extensions that RAMA should treat as model files.
• metamodels: paths of metamodel files used by RAMA when analyzing model files.

No metamodel_extensions option is required, as RAMA supports the two main EMF metamodel extensions: .ecore and .emf.

Use extensions including the leading dot (for example, .model) and repository-relative paths for metamodels:

{
  "model_extensions": [".xmi"],
  "metamodels": ["metamodels/domain.ecore"]
}

Configured metamodels are loaded from the GitHub Actions workspace, which is the target branch when the supplied pull_request_target workflow is used. Consequently, a pull request that adds a model and the metamodel it depends on cannot yet be fully loaded by RAMA: the new metamodel is not available in that workspace. Merge the metamodel first, or expect RAMA to report a loading failure for the model. Loading metamodels directly from the pull-request revision is a planned enhancement.

Automated system testing

RAMA uses reprogit to generate a deterministic Git repository for automated system tests. reprogit builds a repository history from ordered fixture folders, allowing tests to exercise commits, branches, pull requests, and history-dependent workflows.

tests/system/fixture/ contains this test fixture. Its files and commit history are test data: preserve both unless a test intentionally requires a change to the generated repository history. See the reprogit repository (https://github.com/alfonsodelavega/reprogit) for fixture format and usage details.

Acknowledgements

RAMA's development was started as the undergraduate project of Samuel Díaz-Aja at the University of Cantabria. You can access the original project here.

RAMA has been demonstrated at the MODELS'26 tools and demonstrations track. The associated paper and citation details are as follows:

@inproceedings{rama,
    author = {D{\'i}az-Aja, Samuel and S{\'a}nchez, Pablo and de la Vega, Alfonso},
    title = {Facilitating model reviews in pull/change request processes with RAMA},
    year = {2026},
    isbn = {9798400729034},
    url = {https://doi.org/10.1145/3837062.3838869},
    doi = {10.1145/3837062.3838869},
    booktitle = {Proceedings of the ACM/IEEE 29th International Conference on Model Driven Engineering Languages and Systems},
    pages = {204–208},
    numpages = {5},
    series = {MODELS Companion '26}
}