github-pipeline-triage

Automated GitHub issue and PR triage system with heuristic classification, noise detection, duplicate finding, and PR suspicion analysis.

Install

pip install github-pipeline-triage

Usage

# Run full sync and generate ISSUES.md report
github-pipeline-triage sync

# Output as JSON
github-pipeline-triage sync --json

# Skip cache
github-pipeline-triage sync --no-cache

# Skip PR diff analysis
github-pipeline-triage sync --skip-diffs

# Audit suspicious PRs
github-pipeline-triage audit-prs

# Show noise candidates
github-pipeline-triage noise-report

# Start API server
github-pipeline-triage serve --port 8000

API

The server provides REST and WebSocket endpoints:

• GET /api/items - List triage items
• GET /api/stats - Statistics
• GET /api/activity - Activity log
• GET /api/maintainers - Maintainer list
• WS /ws - Real-time updates

Development

git clone https://github.com/dclavijo/github-pipeline-triage.git
cd github-pipeline-triage
pip install -e ".[test]"

# Run tests
pytest

# Format
ruff format src/ tests/

# Lint
ruff check src/ tests/

# Type check
mypy src/

Architecture

• Core: Heuristic classification, noise detection, duplicate finding
• Adapters: GitHub CLI wrappers, caching
• Services: Pipeline, rendering, auditing
• API: FastAPI server with WebSocket
• DB: SQLAlchemy with SQLite