A personal news radar that surfaces relevant articles from HN, RSS, and arXiv based on your long-term interests stored in clawsqlite.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install clawfeedradar
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install clawfeedradar using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is clawfeedradar?
clawfeedradar is an advanced Openclaw Skills extension designed to act as an automated curator for your information diet. By interfacing directly with a clawsqlite knowledge base, it leverages your existing reading history and interest clusters to score new content from the web. It identifies what you are likely to care about by comparing candidate articles against your personal embedding space.
Beyond simple discovery, the skill automates the labor-intensive process of content consumption. It fetches full text, generates AI-powered summaries using small LLMs, and can even produce bilingual body text. The final output is a clean, personalized RSS feed (complete with a detailed JSON sidecar) that can be published to private or public git repositories for easy access across all your devices.
clawfeedradar Use Cases
- Automating daily briefings by filtering Hacker News and arXiv for topics matching your research history.
- Building a self-hosted, personalized newsletter that evolves as your clawsqlite knowledge base grows.
- Generating bilingual news feeds for language learners or cross-border technical research.
- Monitoring specific RSS feeds with an AI filter to eliminate noise and clickbait.
How clawfeedradar Works
- Source Fetching: The skill retrieves raw candidate items from external sources like RSS feeds, Hacker News, or arXiv.
- Deduplication: It normalizes URLs and checks against a 7-day statefile to ensure you never see the same recommendation twice.
- Interest Scoring: Candidates are embedded into a vector space and scored against your clawsqlite interest clusters using a weighted sigmoid function.
- Refinement: A small bias is added based on recency and popularity to ensure the feed remains fresh and high-quality.
- AI Augmentation: For top-tier matches, the skill optionally uses a small LLM to generate summaries and bilingual translations.
- Distribution: Results are written as RSS XML and JSON sidecars to a local directory or automatically pushed to a git repository.
clawfeedradar Setup
To install the skill within your OpenClaw environment, run the following command:
openclaw skills install clawfeedradar
This command initializes the workspace and executes the bootstrap_deps.sh script to install the necessary Python dependencies. Ensure your environment variables for CLAWSQLITE_ROOT and your preferred embedding provider are configured in your agent or host environment.
clawfeedradar Data Schema & Taxonomy
clawfeedradar maintains a strict separation between its output and your primary knowledge base. It organizes data as follows:
| File Type | Purpose | Location |
|---|---|---|
.xml |
Standard RSS feed for reader apps. | CLAWFEEDRADAR_OUTPUT_DIR |
.json |
Metadata sidecar including interest scores and fulltext. | CLAWFEEDRADAR_OUTPUT_DIR |
seen_urls |
Statefile to prevent duplicate processing. | Skill Workspace |
sources.json |
Configuration file for batch processing multiple URLs. | Skill Workspace |
It reads from the interest_clusters and interest_meta tables in your clawsqlite database but never modifies them.
clawfeedradar Advanced Features
- Multi-Source Scheduling: Process dozens of feeds simultaneously using the
schedule_from_sources_jsonaction. - Git Integration: Automated publishing to GitHub or Gitee Pages for seamless feed synchronization.
- Custom Scoring Logic: Fine-tune the influence of recency vs. interest using the
CLAWFEEDRADAR_W_RECENCYandCLAWFEEDRADAR_W_POPULARITYvariables. - Bilingual LLM Pipelines: Support for generating dual-language article bodies for international content monitoring.
SKILL.md
Loading
Related Openclaw Skills
Paid Ads Strategy
A comprehensive strategic framework for planning, allocating, and optimizing paid advertising campaigns across search, social, display, and app channels.
Self-Evolution
An advanced closed-loop system enabling AI agents to learn from mistakes and evolve through automated rule distillation.
Longxia Zhipin
Longxia Zhipin is an intelligent orchestration assistant designed to transform company profiles into executable, one-click AI agent team installations.
Cross-Verify
A sophisticated analytical engine designed to verify facts, detect hidden biases, and evaluate logical integrity in text and data.
Localization Strategy
A comprehensive framework for scaling AI and SaaS products globally through technical i18n implementation and market-specific localization strategies.
Multi-Domain Brand SEO
An advanced brand search optimization workflow designed to structure, differentiate, and rank multiple company domains without keyword cannibalization.
METADATA
Featured*
