oss-issue-scout

Works with GitHub — oss-issue-scout uses GitHub issue and repository data to help developers find worthwhile open-source issues to contribute to.

中文 README

The current version calls the GitHub API, searches open issues, and applies a simple score based on repository activity, issue activity, comments, labels, and related signals. It is currently aimed at junior to intermediate developers who want a faster way to find approachable issues.

Features

• Search GitHub open issues with selectable scoring presets
• Filter by language, label, stars, and update recency
• Skip issues that already have linked PRs
• Recommend only unassigned issues by default
• Skip repositories with fewer than 100 stars by default
• Render results as table, markdown, or json
• No third-party dependencies

Usage

pip install oss-issue-scout
oss-issue-scout search --language python --label "good first issue" --limit 5

Using a GitHub token is recommended. It can be about 3x faster than anonymous search and is less likely to hit rate limits. Set it as an environment variable first:

$env:GITHUB_TOKEN="your_github_token"

oss-issue-scout search --language python --limit 5

This example usually returns results in about 15 seconds.

Web UI

The GitHub Pages version is a static trial version. It calls the GitHub API directly from your browser and applies a lightweight score in the frontend. It does not run the Python backend, so search depth and sorting results may differ significantly from the CLI / local web backend. If you want the more complete web scoring, run the project locally. A future update is planned to sync the Pages version with the backend logic. The optional GitHub token is only sent to api.github.com and is not stored by the page.

Install the optional web dependencies before running the local web UI:

python -m pip install -e ".[web]"

On Windows, start the frontend and backend together with:

.\web\start_web.bat

The script starts the backend API at http://localhost:5000, serves the frontend at http://localhost:8000, and opens http://localhost:8000.

Options

--language            Repository primary language, such as python or c++; default: no language filter
--stars-min           Minimum repository stars; defaults to at least 100
--label               Issue label, such as "good first issue" or "bug"; default: no label filter
--updated-days        Issue updated within the last N days; default: no limit
--repo-updated-days   Repository had issue activity within the last N days; default: no limit
--exclude-repo        Exclude a repository (owner/name) from results; can be repeated; default: none
--limit               Number of results, default 6
--preset              Scoring preset: default, junior, intermediate, senior; default: default
--format              Output format: table, markdown, json; default: table

Examples:

oss-issue-scout search
oss-issue-scout search --language python
oss-issue-scout search --language python --label "help wanted" --stars-min 500 --limit 5
oss-issue-scout search --language rust --format json
oss-issue-scout search --language "C++" --label "good first issue" --repo-updated-days 7
oss-issue-scout search --language c --preset intermediate --limit 10
oss-issue-scout search --language python --exclude-repo django/django --exclude-repo pandas-dev/pandas

Scoring

The current score is intentionally simple. It considers:

• Repository stars: moderately active repos get a boost; very large repos may be penalized
• Issue update recency: recently updated issues get a boost; stale issues are penalized
• Repository issue activity: recent issue activity gets a boost
• Beginner-friendly labels: good first issue / help wanted only add points when the repo has at least 3 open issues with those labels
• Comment count: low discussion volume gets a boost; long discussions are penalized

The search step filters out:

• Closed issues
• Archived repositories
• Issues with linked PRs
• Assigned issues
• Repositories with fewer than 100 stars

Search uses the selected scoring preset. If not specified, it uses the default preset.

Tests

python -m unittest discover

Tests use mocked GitHub responses and do not call the real GitHub API.

Next

This project is still small. If it helps you, please consider giving it a ⭐. Discussions will be opened after the project reaches 128+ ⭐.

If you have suggestions or run into problems, please open an issue.

Future versions will continue to improve recommendation quality and usability.

Thanks Contributors