Advanced Customization

For advanced users, FeedCraft provides an admin dashboard to customize the RSS processing workflow.

Accessing the Dashboard

1. Deploy FeedCraft using Docker (see Quick Start).
2. Open your browser and visit http://YOUR_SERVER_IP:10088.
3. Log in with default credentials:
   • Username: admin
   • Password: adminadmin (Please change the password immediately after logging in)

Search Provider Configuration

To use the Search to RSS feature, you must configure a search provider.

Navigate to Settings > Search Provider in the admin dashboard.

Supported Providers

• LiteLLM / OpenAI Compatible

  • API URL: The search endpoint of your provider (e.g., http://litellm-proxy:4000/v1/search).
  • API Key: Your API key. (Leave empty to keep existing key)
  • Tool Name: The specific function calling tool name if required (e.g., google_search for some agents). The tool name is appended to the API URL (e.g. .../v1/search/google_search).

• SearXNG

  • API URL: The base URL of your SearXNG instance (e.g., http://my-searxng.com). The /search path is automatically appended.
  • Engines: (Optional) Comma-separated list of engines to use (e.g., google,bing).

Tip

You can use the Check Connection button to verify connectivity with your provider before saving.

Dependency Services

The Dependency Services dashboard (Settings > Dependency Services) provides a health check overview of all connected external services.

It monitors the status of:

• SQLite: Database connectivity.
• Redis: Cache service connectivity and latency.
• Browser Provider: Headless browser provider availability (required for browser-based fulltext extraction).
• LLM Service: Connectivity to the configured AI provider.
• Search Provider: Connectivity to the configured search engine.

Use this dashboard to troubleshoot connectivity issues if features like “Enhanced Mode” or “Fulltext Extraction” are failing.

You can use the Check Connection button to verify if FeedCraft can successfully connect to the search provider with the provided credentials.

Note

For monitoring internal Craft dependencies (Recipes, Flows, Atoms), use the Craft Dependencies tool.

Advanced Configuration

Docker Environment Variables

You can configure FeedCraft using environment variables in docker-compose.yml.

• FC_BROWSER_PROVIDER: Browser rendering provider. Supported values: browserless-restful (Browserless REST /content) and cdp (Chrome DevTools Protocol, such as CloakBrowser cloakserve).
• FC_BROWSER_ENDPOINT: Endpoint of the selected browser provider. Required for fulltext-plus and HTML-to-RSS Enhanced Mode.
• FC_PUPPETEER_HTTP_ENDPOINT: Legacy Browserless endpoint alias. Still supported when FC_BROWSER_ENDPOINT is empty.
• FC_BROWSER_TIMEOUT: (Optional) Timeout for a single browser render. Accepts Go duration (60s) or milliseconds (60000). Default: 60s. Keep this at or below the browserless CONNECTION_TIMEOUT.
• FC_BROWSER_MAX_CONCURRENCY: (Optional) Global max concurrent browser render jobs (default: 2). Prevents fulltext-plus from opening too many Chrome sessions at once.
• FC_REDIS_URI: Redis connection address. Used for caching to speed up processing and reduce AI token consumption.
• FC_HTTP_USER_AGENT_FEED: (Optional) Default User-Agent for feed-style outbound requests, such as fetching RSS/XML resources. Search provider requests are temporarily grouped into this same rule.
• FC_HTTP_USER_AGENT_HTML: (Optional) Default User-Agent for HTML page fetches, such as fulltext extraction and the HTML-to-RSS tooling. Note: If the value contains spaces or parentheses, it must be enclosed in quotes.
• FC_LLM_API_KEY: API Key for OpenAI or compatible services (like DeepSeek, Gemini, etc.).
• FC_LLM_API_MODEL: Default model to use (e.g., gemini-pro, gpt-3.5-turbo). Multiple Models Support: You can provide a comma-separated list of models (e.g., gpt-3.5-turbo,gpt-4). FeedCraft will randomly select a model for each request and automatically retry with others if a call fails.
• FC_LLM_API_BASE: API endpoint address. For OpenAI-compatible APIs, usually ends with /v1.
• FC_LLM_API_TYPE: (Optional) openai (default) or ollama.
• FC_LLM_MAX_CONCURRENCY: (Optional) Maximum number of LLM API requests executing across the entire FeedCraft process (default: 3). All feeds, Recipes, AtomCrafts, and other LLM features share this global limit. Article tasks may enter the queue concurrently, but no more than this number of upstream requests execute at once; cache hits do not consume capacity.
• FC_DOMAIN_MAX_CONCURRENCY: (Optional) Maximum concurrent requests per target domain during web scraping like fulltext extraction (default: 3). Prevents overwhelming target servers.
• FC_PREHEATING_MAX_CONCURRENCY: (Optional) Maximum concurrent background preheating tasks (default: 2).
• FC_PREHEATING_QUEUE_SIZE: (Optional) Maximum pending preheating tasks; defaults to the preheating concurrency.
• FC_PREHEATING_TASK_TIMEOUT: (Optional) Timeout for each preheating task as a Go duration such as 5m (default: 10m).
• LOG_LEVEL: (Optional) Log level for the backend application (e.g., info, debug, trace). Overrides the default level set by ENV.

External Services

To leverage the full power of FeedCraft, it is recommended to deploy with Redis and a browser provider.

version: "3"
services:
  app.feed-craft:
    # ... (Refer to Quick Start)
    environment:
      FC_BROWSER_PROVIDER: browserless-restful
      FC_BROWSER_ENDPOINT: http://service.browserless:3000
      FC_REDIS_URI: redis://service.redis:6379/
      # ...

  service.redis:
    image: redis:6-alpine
    container_name: feedcraft_redis
    restart: always

  service.browserless:
    image: browserless/chrome
    container_name: feedcraft_browserless
    environment:
      USE_CHROME_STABLE: true
    restart: unless-stopped

To use CloakBrowser instead of Browserless, point FeedCraft to the official cloakserve container:

services:
  app.feed-craft:
    environment:
      FC_BROWSER_PROVIDER: cdp
      FC_BROWSER_ENDPOINT: http://service.cloakbrowser:9222?fingerprint=feedcraft

  service.cloakbrowser:
    image: cloakhq/cloakbrowser
    command: cloakserve --port=9222
    restart: unless-stopped

The service listens on port 80 by default. You can also access it from other containers in the same network using http://app.feed-craft/xxx (e.g., for internal communication with an RSS reader).