project
Horizon - An open-source AI information aggregation system for building your own news radar.
Horizon is an open-source, AI-driven information aggregation system that helps users build their own AI news radar. The tool supports crawling information from multiple sources, including RSS, Hacker News, GitHub, Reddit, Telegram, and Twitter/X...
What is Horizon?
Horizon is an open-source, AI-driven information aggregation system that helps users build their own AI news radar. The tool supports scraping information from multiple sources, including RSS, Hacker News, GitHub, Reddit, Telegram, and Twitter/X. It uses AI for deduplication, scoring, summarizing, and adding background information, ultimately outputting daily bilingual (Chinese and English) briefings in webpage, email, webhook, or MCP formats, allowing users to efficiently access core developments in the AI field.
Horizon's main features
-
Multi-source information aggregationSupports automatic content scraping from channels such as RSS, Hacker News, GitHub Trending, Reddit, Telegram, Twitter/X, and OpenBB.
-
AI intelligent scoringThe system calls upon a large model to score the relevance of information (0-10 points) and filters it into three levels: high signal, referenceable, and low signal.
-
Automatic deduplication and summarizationIt uses AI to remove duplicate content and generate structured summaries, background knowledge, and community discussion extracts.
-
Multi-format outputSupports the generation of web dashboards, email briefings, webhook push notifications, and MCP protocol distribution.
-
Flexible filtering and searchingContent can be filtered by score threshold, information source classification, time window (24h/3 days/7 days), and keyword precise matching.
Horizon's technical principles
-
Asynchronous crawling engineThe asynchronous architecture is used to pull data from multiple information sources in parallel, improving collection efficiency and real-time performance.
-
LLM Scoring and SummaryLarge models such as Claude, GPT, Gemini, DeepSeek, Doubao, MiniMax, OpenClaw, and Ollam are used for content quality assessment and text generation.
-
Deduplication algorithm: Filter duplicate information through content similarity detection mechanism to avoid information redundancy.
-
Background knowledge enhancement: Use AI to supplement background information, organize references, and aggregate community viewpoints for high-value content.
-
Modular output pipelineThe processed content will be distributed through standardized interfaces such as Pages, Email, Webhooks, and MCP.
How to use Horizon
-
Cloning repositoryPull the Horizon project source code from GitHub to your local server.
-
Install dependenciesUse the uv tool to install the Python dependencies required for the project.
-
Configure API KeyEnter the API Key of the large model vendor (such as OpenRouter) and the required information source credentials in the configuration file.
-
Set information sourceEnable crawling channels such as RSS, GitHub, Hacker News, Twitter/X as needed and configure parameters.
-
Start AggregationRun the main program, and the system will automatically capture, score, summarize, and generate a daily briefing.
-
View outputReceive AI-filtered news briefings via generated web dashboards or configured emails/webhooks.
Horizon's core advantages
-
Fully open sourceUsers can freely deploy and customize information sources and output rules, giving them a high degree of data autonomy.
-
Multi-model compatibilityIt supports integration with mainstream and open-source models such as OpenRouter, DeepSeek, and xAI, with controllable and flexible costs.
-
Bilingual briefingAutomatically generates daily briefings in both Chinese and English, covering both domestic and international AI community updates.
-
Low-threshold deploymentIt only requires a large model API Key to run, is easy to configure, and supports quick UV installation.
Horizon's project address
- Project official website: https://www.horizon1123.top/
- GitHub repositoryhttps://github.com/Thysrael/Horizon
Comparison of Horizon's similar products
| Comparison Dimensions | Horizon | AIHOT |
|---|---|---|
| Product Positioning | Open-source AI news radar, allowing users to build their own personalized information streams. | An AI industry hotspot monitoring platform, the "Toutiao of the AI world". |
| Open source license | MIT is completely open source and can be freely modified and developed. | Closed-source commercial products, only allowing access via Skill/API/RSS. |
| Deployment method | Local/server self-hosting, complete data control | Officially hosted (aihot.virxact.com), ready to use immediately. |
| Number of information sources | 7 main categories that can be customized (RSS/HN/GitHub/Reddit/TG/X/OpenBB) | 168 carefully selected sources (RSS/Twitter/Weibo/GitHub/Papers/Official Blog) |
| Source strategy | User-configurable and flexibly expandable | The Kha'Zix team manually selects and maintains the products, ensuring quality. |
| AI screening mechanism | Single-level AI scoring (0-10 points) + deduplication + summary + background knowledge supplementation | Two-stage AI filtering: DeepSeek V3.2 pre-screening and noise reduction + V4 Pro high-quality scoring. |
| Content Classification | Categorized by source (GitHub/HN/RSS/X, etc.) | Categorized by topic (Model/Product/Industry/Paper/Skills) |
| Output format | Web dashboards, email briefings, webhooks, MCP protocol | Web browsing, RSS subscription, REST API, Agent Skill |
| Agent Access | Supports MCP protocol output | Provides a standard SKILL.md file, compatible with agent platforms such as Claude, Codex, and Cursor. |
| Model support | Multi-model compatibility (Claude/GPT/Gemini/DeepSeek/Doubao/MiniMax/Ollama, etc.) | The backend consistently uses DeepSeek V3.2 + V4 Pro. |
| Language support | Daily Briefing in Chinese and English | Chinese as the main language |
| Usage cost | Only incur API Key call fees. | Completely free, no API key required |
Horizon Application Scenarios
-
Daily News Roundup for AI PractitionersQuickly stay up-to-date with popular GitHub projects, Hacker News technical discussions, and official Twitter updates.
-
Tech self-media content selectionHigh-value topics are selected through scoring to help determine article topics and track trending issues.
-
Team Technical Radar ConstructionIt aggregates cutting-edge technology trends for enterprise technology teams and regularly pushes internal briefings.
-
Personal knowledge management: Build customized information streams to replace traditional RSS readers and achieve AI-assisted filtering.