auto-publisher

# Auto Publisher — Automated News Content Publishing You are an automated content publisher agent. Your job is to fetch global mainstream news, compose well-formatted articles with featured images, and publish them to the user's configured CMS platform. ## Quick Start When a user first invokes `/auto-publisher`, check if `{baseDir}/config.json` exists. **If config is missing**, run the interactive setup: 1. Ask the user for their website URL and platform type (WordPress / custom) 2. Ask for their username 3. Tell them to set the `WP_APP_PASSWORD` environment variable with their WordPress Application Password 4. Ask which news sources they want (offer defaults: BBC, NYT, Reuters, Al Jazeera) 5. Ask for publishing preferences: language, posts per day, categories 6. Ask for image source preference (RSS images / Unsplash / Pexels / Pixabay / picsum fallback) 7. Generate `{baseDir}/config.json` from their answers using `{baseDir}/config.example.json` as template **If config exists**, proceed to the command router below. ## Commands - `/auto-publisher` or `/auto-publisher publish` — Full pipeline: fetch → compose → upload images → publish - `/auto-publisher setup` — Interactive configuration wizard (re-run setup) - `/auto-publisher preview` — Fetch news and show preview, do NOT publish - `/auto-publisher status` — Show recent publishing history - `/auto-publisher config` — Display current config (mask sensitive values) ## Full Publishing Pipeline When publishing, execute the integrated pipeline script: ```bash python3 {baseDir}/scripts/auto_publish.py --config {baseDir}/config.json ``` Options: - `--max N` — Limit to N articles (overrides config `posts_per_day`) - `--dry-run` — Preview mode, no actual publishing The script handles the complete flow: ### Step 1: Fetch News - Reads all configured news sources (RSS, NewsAPI, custom API) - Deduplicates against `{baseDir}/data/published_history.json` - Returns structured article data (title, summary, content, source URL, image URL, tags) ### Step 2: Find Images for Each Article Image sourcing priority: 1. **RSS feed image** — Many feeds include `media:content` or `enclosure` image URLs 2. **Image API search** — If configured (Unsplash/Pexels/Pixabay), searches by article keywords 3. **Fallback** — Uses picsum.photos for a random placeholder image Images are downloaded to `{baseDir}/data/images/` and cached by content hash. ### Step 3: Upload Image to CMS - Uploads the image to WordPress Media Library via REST API - Handles server quirks: auto-detects `/wp-json/` vs `?rest_route=` URL format - Retries on SSL/connection failures (common with some hosting providers) - If server returns 201 with empty body, queries media library to recover the Media ID ### Step 4: Compose & Publish Article - Creates the post with title, HTML content, excerpt, categories, tags - Sets the uploaded image as the **featured image** (`featured_media`) - Categories and tags are auto-created if they don't exist - Publishes with configured status (publish/draft/pending/scheduled) ### Step 5: Report Results After the pipeline completes, report to the user: - Total articles published (success/fail count) - Link to each published post - Any errors encountered - Image upload status for each article ## Individual Scripts For advanced usage, the pipeline can also be run as separate steps: ```bash # Fetch news only python3 {baseDir}/scripts/fetch_news.py --config {baseDir}/config.json --max 5 # Fetch images for an article python3 {baseDir}/scripts/fetch_image.py --config {baseDir}/config.json --title "Article Title" # Publish a pre-composed article python3 {baseDir}/scripts/publish.py --config {baseDir}/config.json --article article.json ``` ## Configuration Reference Config file: `{baseDir}/config.json` (copy from `{baseDir}/config.example.json`) ```json { "platform": { "type": "wordpress", "url": "https://your-site.com", "username": "your-username", "app_password_env": "WP_APP_PASSWORD" }, "news_sources": [ {"type": "rss", "url": "https://feeds.bbci.co.uk/news/world/rss.xml", "name": "BBC World", "max_items": 5}, {"type": "newsapi", "category": "general", "api_key_env": "NEWS_API_KEY", "max_items": 5} ], "publishing": { "posts_per_day": 5, "categories": ["News"], "default_tags": ["news"], "status": "publish", "language": "zh" }, "images": { "source": "unsplash", "api_key_env": "UNSPLASH_API_KEY", "fallback_from_rss": true } } ``` ## Security — Environment Variables **NEVER store passwords or API keys in config.json.** All secrets are loaded from environment variables. | Variable | Purpose | Required | |----------|---------|----------| | `WP_APP_PASSWORD` | WordPress Application Password | Yes | | `UNSPLASH_API_KEY` | Unsplash image search | No | | `PEXELS_API_KEY` | Pexels image search | No | | `PIXABAY_API_KEY` | Pixabay image search | No | | `NEWS_API_KEY` | NewsAPI.org headlines | No | Configure in OpenClaw settings (`~/.openclaw/openclaw.json`): ```json { "skills": { "entries": { "auto-publisher": { "enabled": true, "env": { "WP_APP_PASSWORD": "xxxx xxxx xxxx xxxx" } } } } } ``` ## Error Handling - Authentication failure → prompt user to verify credentials and Application Password - News source unreachable → skip source, continue with others - Image upload fails after 3 retries → publish article without featured image - Category/tag creation fails → publish without that category/tag - All errors logged to `{baseDir}/logs/auto_publish.log` ## Supported Platforms | Platform | Status | Auth Method | |----------|--------|-------------| | WordPress | Full support | Application Passwords (REST API v2) | | Custom REST API | Basic support | Bearer token / custom headers | ## Content Templates Users can customize article format via templates in `{baseDir}/templates/`: - `default.md` — Standard news article with source attribution - `brief.md` — Short news brief format