For heavy RSS users, the worst experience is clicking on a feed they're interested in, only to have the page display only... Title and brief summaryThis forces users to jump to the original webpage filled with ads to read the full text. Furthermore, the cost of filtering through a massive amount of long English articles or mixed marketing content is extremely high.
FeedCraft This is an open-source, self-hosted RSS middleware designed to address the aforementioned pain points. It doesn't function directly as a reader, but rather plays the role of "data cleaning and reconstruction." By deploying it between the RSS feed and the reader, it can process the "semi-finished" digest feed into... Includes full-text extraction, intelligent translation, and AI summarization. A high-value source of in-depth reading.
How it works: From "tableware" to "kitchen"
To better understand FeedCraft's positioning, it can be compared with traditional e-readers:
- RSS readers (such as Reeder, Feedly): It is equivalent to "tableware," responsible for the final presentation and consumption of content.
- FeedCraft (middleware): It's like the "kitchen," responsible for cleaning, extracting, and translating (cooking) the raw RSS data (raw material) before delivering it to the reader.
This means you don't need to change your preferred reading software; you can simply replace the subscription address with the new URL processed by FeedCraft.
Core functional modules (AtomCraft)
FeedCraft via built-in AtomCraft The processing module allows users to customize the RSS processing flow like building blocks.
1. Force Fulltext Extraction
For sites that only provide summaries, FeedCraft can automatically crawl the full text. For dynamic web pages that require JavaScript rendering, it supports simulating browser rendering, restoring a "crippled source" to a "full-text source".
2. AI-powered immersive translation (Translate)
Supports integration DeepSeek, OpenAI, Ollam For mainstream interfaces, two translation modes are provided:
- Title Translation: Only the title is translated for easy and quick filtering.
- Immersive comparison: It adopts a bilingual (Chinese and English) format, balancing reading efficiency with language learning.
3. Intelligent Noise Reduction and Summarization (Filter & Summary)
Generate using AI TL;DR (Too long to read) The summary is pinned to the top to help users quickly grasp the key points. It also automatically blocks marketing content containing words like "promotion" or "special offer" through keyword rules.
4. Everything can be RSS (Any to RSS)
For websites that do not have a native RSS interface, FeedCraft can... Web page HTML, API interface Even search engine results can be directly converted into subscribed RSS feeds.
Applicable Scenarios
Beyond personal reading, FeedCraft has even greater value in enterprise-level information automation:
- Industry intelligence monitoring: It automatically captures competitor updates and technology community updates, and generates daily AI briefings.
- High-purity information flow: Filter out redundant information from the media and retain only the core value content.
- Knowledge base automation: The processed feed can be integrated into Notion, Slack, or Lark to build an automatically updated research resource pool.
- Private deployment: The data is completely controlled on our own servers, meeting the company's security and compliance requirements.
Competitive product comparison
| Tool type | Core pain points | Advantages of FeedCraft |
|---|---|---|
| Traditional e-reader | Missing content, excessive advertising | Source purificationWash first, then read. |
| RSSHub | It focuses on the "generation" source and lacks in-depth processing. | Deep processingIt can be used in conjunction with RSSHub. |
| Read Later Software | Single-article manual storage is not feasible for batch automation. | Assembly line operationAutomated processing of the entire feed |
Precautions
- API cost control: Full text translation will consume a large amount of tokens. We recommend prioritizing [other methods]. DeepSeek V3 or Gemini Flash Low-cost models, or local deployment Ollama。
- Hardware resources: Enabling "Fulltext Plus" will significantly increase memory usage; please use caution when enabling it on low-configuration VPS.
- Safety reinforcement: After self-hosting deployment, please be sure to... Change the default backend passwordTo avoid API key leakage.
How to obtain
📦 GitHub Project Homepage
Includes Docker Compose deployment tutorials and API documentation.
🎮 Experience the Demo Online
Deployment-free testing of full-text extraction and translation results
⚠️ Disclaimer: This project is maintained by the open-source community. Please comply with the Robots Exclusion Protocol of the target website and set a reasonable crawling frequency to avoid putting excessive pressure on the target site.
