# dnr-automation

> Automate the Daily News Roundup (DNR) newsletter pipeline for the Center for Cooperative Media. Run the DNR script with different modes (--preview generates HTML preview only, --dry-run shows story counts without generating files, full pipeline creates Mailchimp draft). Use this skill when working on DNR newsletter automation, running the pipeline, previewing newsletters, testing story collection and classification, or troubleshooting the NJ News Commons newsletter workflow.

- Author: Joe Amditis
- Repository: CenterCoopMedia/dnr
- Version: 20251225171732
- Stars: 0
- Forks: 0
- Last Updated: 2026-02-06
- Source: https://github.com/CenterCoopMedia/dnr
- Web: https://mule.run/skillshub/@@CenterCoopMedia/dnr~dnr-automation:20251225171732

---

---
name: dnr-automation
description: Automate the Daily News Roundup (DNR) newsletter pipeline for the Center for Cooperative Media. Run the DNR script with different modes (--preview generates HTML preview only, --dry-run shows story counts without generating files, full pipeline creates Mailchimp draft). Use this skill when working on DNR newsletter automation, running the pipeline, previewing newsletters, testing story collection and classification, or troubleshooting the NJ News Commons newsletter workflow.
allowed-tools: Bash, Read, Glob, Grep
---

# DNR Newsletter Automation

## When to Activate

Activate this skill when:
- User wants to run the DNR newsletter pipeline
- User mentions "create newsletter", "run DNR", or "generate preview"
- Troubleshooting pipeline issues
- Testing story collection or classification
- Checking story counts or feed connectivity

## Core Concepts

**The 7-Step Workflow**:
```
1. Fetch → RSS (75+) + Airtable + optional Playwright
2. Enrich → Optional Gemini API summaries
3. Classify → Claude Haiku assigns sections
4. Review → Approve Airtable submissions (triggers email)
5. Preview → Generate HTML, open in browser
6. Refine → Natural language feedback loop
7. Publish → Create Mailchimp draft
```

**Date-Aware Fetching**:
| Day | Lookback | Reason |
|-----|----------|--------|
| Monday | 76 hours | Covers Fri 5am - now (weekend) |
| Tue-Thu | 36 hours | Standard daily coverage |
| Fri-Sun | Warning | Not normal publish days |

**Execution Modes**:
| Flag | Purpose |
|------|---------|
| (none) | Full pipeline with Mailchimp draft |
| --preview | Generate HTML only, no Mailchimp |
| --dry-run | Story counts only, no output |
| --playwright | Include Playwright scraping |
| --enrich | Add Gemini URL enrichment |

## Practical Guidance

### Quick Start Commands

```bash
cd /home/user/dnr

# Standard workflow (recommended)
venv/Scripts/python.exe src/workflow.py

# With Playwright for paywalled sites
venv/Scripts/python.exe src/workflow.py --playwright

# Preview only (testing)
venv/Scripts/python.exe src/main.py --preview

# Quick story count check
venv/Scripts/python.exe src/main.py --dry-run

# Test API connections
venv/Scripts/python.exe src/test_connections.py
```

### Newsletter Sections

| Section | Internal Name | Story Count |
|---------|---------------|-------------|
| Top stories | top_stories | 3-6 max |
| Politics + government | politics | 5-15 |
| Housing + development | housing | 3-10 |
| Work + education | education | 3-10 |
| Health + safety | health | 3-10 |
| Climate + environment | environment | 3-10 |
| Lastly | lastly | 5-20 |

### Troubleshooting

**Pipeline fails with API errors:**
- Run `src/test_connections.py` to verify credentials
- Check `.env` file has all required keys

**Low story count:**
- Try `--hours 48` for broader window
- Check RSS feed health
- Run with `--playwright` for paywalled sites

**Classification issues:**
- Review `src/classifier.py` section descriptions
- Use feedback loop in Step 6 to move stories

## Examples

**Input**: "Run the newsletter"

**Output**: Execute standard workflow:
```bash
cd /home/user/dnr
venv/Scripts/python.exe src/workflow.py
```

**Input**: "Just preview, don't create Mailchimp draft"

**Output**: Run preview mode:
```bash
venv/Scripts/python.exe src/main.py --preview
```

**Input**: "How many stories do we have today?"

**Output**: Run dry-run:
```bash
venv/Scripts/python.exe src/main.py --dry-run
```

## Guidelines

1. Always run from /home/user/dnr directory
2. Use workflow.py for interactive production runs
3. Use main.py with flags for automated/testing runs
4. Check test_connections.py if APIs fail
5. Monday newsletters need full weekend coverage

## Integration

- **dnr-quality-audit**: Audit after preview generation (Step 5)
- **dnr-editorial-feedback**: Process feedback during Step 6
- **dnr-feed-health**: Diagnose if stories are missing
- **dnr-airtable-triage**: Speed up Step 4 review

## File References

- Workflow script: `src/workflow.py`
- Pipeline script: `src/main.py`
- Connection test: `src/test_connections.py`
- Output directory: `drafts/`
- Template: `history/dnr-template.html`

## Environment Requirements

```
ANTHROPIC_API_KEY=     # Claude Haiku classification
AIRTABLE_PAT=          # Airtable submissions
AIRTABLE_BASE_ID=
AIRTABLE_TABLE_ID=
MAILCHIMP_API_KEY=     # Campaign creation
MAILCHIMP_SERVER_PREFIX=
MAILCHIMP_LIST_ID=
GEMINI_API_KEY=        # Optional: URL enrichment
```