hubsignal

Find promising open-source work on GitHub — in seconds, not hours.

hubsignal is a zero-dependency CLI that scans GitHub issues and ranks them by signal strength. Stars, labels, freshness, and discussion activity all factor into a simple, inspectable score.

Quick Start

git clone https://github.com/Photon101/hubsignal.git
cd hubsignal
uv run python -m hubsignal --limit 10

What You Get

 score    stars  repo                              issue
────────────────────────────────────────────────────────────────────────────
  84.2      824  openjournals/joss-reviews         [REVIEW]: MaterForge: Materials Formulation Engine
                                                   https://github.com/openjournals/joss-reviews/issues/9909

  72.1    20473  microsoft/onnxruntime             Handle empty initializers gracefully in optimizer passes
                                                   https://github.com/microsoft/onnxruntime/pull/27976

  58.8      120  Consensys/linea-monorepo          [ZkTracer] Add test cases for blake.rs implementation
                                                   https://github.com/Consensys/linea-monorepo/pull/3011

Each issue gets a score from 0–100. Higher scores mean the issue is in an active, starred repo with useful labels and recent activity — the kind of issue where your contribution matters.

Features

Flag	What it does
--min-stars N	Only show repos with ≥N stars
--exclude-archived	Skip archived repositories
--exclude-forks	Skip forked repositories
--exclude-bounty-like	Filter out token promotions, star-for-reward, crypto noise
--bounty-only	Oppose of above — only show issues with bounties/rewards
--exclude-repo OWNER/NAME	Exclude specific repos (repeatable)
--exclude-title-regex PATTERN	Filter by title regex
--pushed-after YYYY-MM-DD	Only repos active after this date
--format text|json|markdown	Output as plain text, JSON, or Markdown report
--limit N	Number of issues to scan (1–100)

Real Use Cases

Finding your first open-source contribution:

hubsignal --query 'is:issue is:open label:"good first issue" label:documentation' \
  --min-stars 20 --exclude-archived --exclude-forks --exclude-bounty-like

Hunting paid bounties:

hubsignal --bounty-only --query 'label:bounty is:issue is:open' --limit 20

Weekly planning report:

hubsignal --limit 30 --format markdown > weekly-scout-$(date +%Y-%m-%d).md

JSON export for your own tooling:

hubsignal --format json | jq '.issues[] | {repo, url, score}'

How Scoring Works

score = 20 (base)
      + min(35, log10(stars + 1) × 9)   ← repo popularity
      + max(0, 18 − age_days × 0.35)     ← issue freshness
      + min(10, log2(comments + 1) × 2)  ← discussion activity
      + 10 (if "help wanted" label)
      + 6  (if "good first issue" label)
      + 12 (if bounty/reward label)
      + 3  (if bug label)

Fully inspectable. No hidden weights. No AI. No API keys needed.

Authentication

GitHub limits unauthenticated API requests. For best results:

export GH_TOKEN=your_github_token
# or install gh CLI and login:
gh auth login

hubsignal checks GH_TOKEN, GITHUB_TOKEN, then gh auth token — in that order.

Install

git clone https://github.com/Photon101/hubsignal.git
cd hubsignal
uv run python -m hubsignal --help

Zero dependencies at runtime. pytest is optional for development.

Why I Built This

I spent hours manually scrolling GitHub issue search looking for good first contributions. The signal-to-noise ratio is terrible — most issues are stale, claimed, or in dead repos. hubsignal cuts that first hour of scanning into a 30-second command.

Sponsor

If hubsignal saves you time finding open-source work, consider sponsoring on GitHub. Sponsorship funds sustained development and keeps the tool dependency-free and focused.

License

MIT