Kagi

What This Skill Does

The Kagi skill for OpenClaw integrates the premium Kagi search and FastGPT services directly into your agentic workflow. Unlike standard free search APIs, Kagi provides high-quality, ad-free, and privacy-focused search results. The skill provides two distinct modes: a traditional web search capability that returns a ranked list of URLs and snippets, and a sophisticated FastGPT service that processes natural language queries to deliver concise, citation-backed summaries of current web information. By leveraging the kagi_client.py library, the skill ensures that your agent can fetch real-time data effectively, acting as a powerful alternative when other search providers face rate limits or fail to provide sufficient context.

Installation

To begin, ensure you have an active account at Kagi and have generated an API token from your account settings. Installation is handled via the OpenClaw CLI tool. Run the command clawhub install openclaw/skills/skills/michaelasper/kagi within your terminal. After installation, you must configure your environment by exporting your unique token: export KAGI_API_TOKEN='your_token_here'. This allows the python scripts in the scripts/ directory to authenticate correctly with the Kagi API endpoints.

Use Cases

This skill is ideal for research-intensive tasks where result accuracy and noise reduction are paramount. Use the Search API when your agent needs to compile a list of source materials or verify technical documentation. Use the FastGPT function for rapid knowledge synthesis, such as summarizing market news, explaining complex concepts using up-to-date web data, or answering questions that require grounding in recent online discourse. It acts as a primary research engine for agents tasked with report generation or competitive analysis.

Example Prompts

1. "Search with Kagi for the latest updates on the OpenClaw framework and provide a list of relevant documentation URLs."
2. "Kagi FastGPT: Summarize the primary differences between current AI agent architectures based on this week's technical blog posts."
3. "Use Kagi to find the most recent documentation for the Haaps glass manufacturing process and summarize the main technical hurdles."

Tips & Limitations

Kagi operates on a paid API token model, so monitor your usage to avoid unnecessary costs. If you receive a 429 rate-limit error from standard search tools like Brave, immediately switch to the Kagi skill for consistent performance. Remember that while FastGPT is excellent for summarization, always verify sensitive or critical information against the provided source URLs. If you require raw data for secondary processing, always include the --json flag in your scripts to receive machine-readable output rather than human-readable text digests.