Feature: Build a NEWS Bot for Lecture Repositories

[mmcky]

Summary

Build a NEWS bot (GitHub Action) that monitors QuantEcon lecture repositories for recent changes and automatically opens a PR with a suggested news summary for the website or newsletter.

Motivation

Keeping the QuantEcon community informed about updates to our lecture content is important but manual. A bot that automatically detects significant changes and drafts news summaries would:

• Reduce manual effort in tracking changes across multiple repositories
• Ensure timely communication of updates to our audience
• Provide consistent formatting for news items
• Make it easier to maintain an active news/changelog presence

Proposed Functionality

Core Features

1. Monitor Lecture Repositories - Scan configured lecture repositories for recent commits/PRs merged
2. Detect Significant Changes - Filter for meaningful updates (new lectures, major revisions, bug fixes)
3. Generate News Summary - Use AI to draft a human-readable summary of changes
4. Open PR with Summary - Automatically create a PR with the suggested news content

Configuration Options

• List of repositories to monitor
• Frequency of checks (weekly, bi-weekly)
• Types of changes to include (new content, updates, fixes)
• Output format (markdown, newsletter format)
• Target repository for news PRs

Example Workflow

name: Generate News Summary
on:
  schedule:
    - cron: '0 9 * * 1'  # Weekly on Monday
  workflow_dispatch:

jobs:
  news-bot:
    runs-on: ubuntu-latest
    steps:
      - uses: QuantEcon/action-news-bot@main
        with:
          repositories: |
            QuantEcon/lecture-python-intro
            QuantEcon/lecture-python.myst
            QuantEcon/lecture-jax
          lookback-days: 7
          ai-summary: true
          target-repo: QuantEcon/website
          target-branch: news-updates

Implementation Considerations

• Should integrate with existing QuantEcon action patterns
• Could leverage the same AI infrastructure as action-link-checker
• News summaries should be editable before final publication
• Consider rate limiting and API usage for AI summarization

Related

• action-weekly-report - Similar concept for issue/PR activity reports
• action-link-checker - Reference for AI-powered actions

Tasks

• Design news summary format/template
• Implement repository change detection
• Integrate AI summarization
• Build PR creation workflow
• Add configuration options
• Documentation and examples

[mmcky]

Status: this shipped — and running it for a month produced a design lesson that's now driving the next phase.

The bot exists as the Daily News Watch / weekly news workflows in QuantEcon/reports-activity (private), publishing via the quantecon-news-reporter App as review-gated draft PRs to QuantEcon/website (e.g. QuantEcon/website#241).

The lesson: generated release/update digests dilute the curated news stream, and a human review gate on deterministically-generated content adds burden without value. The proposed evolution splits quantecon.org news (rare, hand-written announcements) from a new auto-published, tag-filterable Activity feed. Design is tracked in QuantEcon/project-reporting#4 (Phase A, current Jekyll stack) and QuantEcon/project-reporting#5 (Phase B, longer-term stack rebuild with an adversarial stack evaluation) — private while in proposal stage.