| name | rank-changes |
| description | Evaluate GitHub changes for community interest and produce a ranked list |
Rank Changes for Community Digest
You are given a list of recent GitHub changes (commits, merged PRs, new/closed issues) from a repository.
Task
Evaluate each change for community interest — how relevant and interesting is this change to developers and domain experts following the project?
Output Format
Write a JSON file to /output/result.json with this structure:
{
"output_file": "/output/result.json",
"summary": "Ranked N changes from REPO"
}
The result.json should contain:
{
"rankedChanges": [
{
"rank": 1,
"score": 9,
"category": "feature",
"title": "Short title of the change",
"description": "One-line rationale for the ranking",
"source": "pr",
"sourceId": "123",
"url": "https://github.com/..."
}
],
"metadata": {
"totalChanges": 42,
"rankedCount": 10,
"topCategory": "feature"
}
}
Scoring Guidelines
- Score 1-3: Internal/infra changes (CI fixes, dependency bumps, typo fixes)
- Score 4-6: Meaningful but routine (bug fixes, minor improvements, docs updates)
- Score 7-9: High community interest (new features, breaking changes, major fixes, architectural decisions)
- Score 10: Rare — milestone releases, major new capabilities
Categories
feature — new functionality
bugfix — bug fix
infra — CI/CD, tooling, dependencies
docs — documentation changes
refactor — code restructuring without behavior change
breaking — breaking changes
Rules
- Rank the top 10-15 changes maximum (skip the rest)
- Be ruthless — most CI/infra changes are score 1-2
- Features and breaking changes almost always rank higher than fixes
- If a PR has a good description, use it; if not, infer from the diff