searxng

What This Skill Does

The SearXNG skill for OpenClaw agents provides a powerful, privacy-focused web search capability without the need for external API keys. SearXNG acts as a self-hosted search aggregator, meaning it queries multiple search engines (such as Google, Bing, Brave, Startpage, DuckDuckGo, and Wikipedia) on your behalf and consolidates the results. This skill offers two primary functionalities: it can install and configure a SearXNG instance on a Virtual Private Server (VPS) or your own server, making web search available to your agent, or it can connect to an already existing local SearXNG instance. The skill covers the entire lifecycle from initial installation, configuration, usage via a command-line interface tool (search.py), and even includes fallback mechanisms to ensure search functionality even if the SearXNG instance is temporarily unavailable.

Installation

For users setting up a new OpenClaw instance or requiring a dedicated web search backend, the skill provides an automated installation script. On Ubuntu 22.04 or 24.04, run the following command as root:

bash scripts/install_searxng.sh

This script automates the installation of SearXNG, creates a dedicated searxng system user, configures the necessary settings in /etc/searxng/settings.yml, and sets up a systemd service to manage the SearXNG process. By default, SearXNG will be accessible at http://127.0.0.1:8888. You can verify the installation by checking the service status with systemctl status searxng and by fetching test results using curl.

If you already have a SearXNG instance running, ensure that the search.py script (typically located in tools/search.py within your workspace) is configured to point to the correct URL and port. The script defaults to http://127.0.0.1:8888, but you can modify the SEARXNG_URL variable within the script if your instance runs on a different address or port.

Use Cases

This skill is invaluable for OpenClaw agents that require up-to-date information from the web. Key use cases include:

1. Web Information Gathering: Agents can perform general web searches to gather information on any topic, retrieve current news, research products, or find factual data.
2. New OpenClaw Instance Setup: When deploying a new OpenClaw instance, integrating SearXNG ensures immediate access to web search capabilities without relying on potentially rate-limited or API-key-dependent services.
3. Troubleshooting Search Failures: If your agent experiences issues with web searches, this skill provides diagnostic steps and fallback information to identify and resolve the problem, whether it's a configuration issue with SearXNG or a network problem.
4. Privacy-Conscious Searching: For users who prefer not to rely on large search providers or want to avoid tracking, SearXNG's self-hosted nature offers a more private alternative.

Example Prompts

1. "Find the latest research papers on quantum computing published in the last month."
2. "What are the current stock prices for NVIDIA and AMD?"
3. "Summarize the key points from the recent G7 summit."

Tips & Limitations

• Resource Intensive: Running your own SearXNG instance requires server resources (CPU, RAM, bandwidth). Ensure your VPS or server is adequately provisioned.
• Configuration: While the installation script automates much of the setup, advanced customization of SearXNG (e.g., enabling specific search engines, fine-tuning results) requires editing the /etc/searxng/settings.yml file.
• Fallback Behavior: In the event your SearXNG instance is down, the search.py script intelligently falls back to using Wikipedia and GitHub APIs. While this ensures continued search functionality, the scope of results will be narrower.
• CLI Tool: The search.py script is the primary interface for performing searches. Use the --json flag for machine-readable output that can be easily parsed by your agent, and the --count flag to limit the number of results.