Web Fetch
The Web Fetch tool retrieves the full content of a web page and makes it available to the AI for reading and analysis. It goes beyond search snippets — the AI gets the complete text of the page.
What It Does
Section titled “What It Does”- Fetches the full rendered content of a URL
- Strips navigation, ads, and boilerplate — returns the main content
- Supports articles, documentation, GitHub pages, and most public web content
Fetch Methods
Section titled “Fetch Methods”ChatShell supports multiple methods for fetching web content:
| Method | Description |
|---|---|
| Local (Readability + Headless Chrome) | Uses Mozilla’s Readability library combined with headless Chrome for full JavaScript rendering. Best for complex pages but requires more resources. |
| Local (Readability Only) | Fast extraction using Readability alone. Good for articles and blog posts. |
| Local (Headless Chrome Only) | Full browser rendering without content extraction. Useful for SPAs and dynamic content. |
| Jina Reader API | Sends URL to Jina’s Reader API and returns clean, structured content. Fast and reliable but requires internet access to external API. |
Configure the default fetch method in Settings → Tools → Web Fetch. You can also choose “Auto” to let ChatShell decide the best method based on the URL.
When It’s Used
Section titled “When It’s Used”The AI uses Web Fetch when:
- A URL is shared directly in the conversation
- Web Search results need to be read in depth
- Documentation or reference material needs to be analyzed
Example
Section titled “Example”“Read the Tokio docs getting started page and summarize the key concepts.”
The AI will use Web Fetch on the Tokio documentation URL and produce a structured summary of the content.
Limitations
Section titled “Limitations”- JavaScript-heavy single-page apps may not render correctly
- Pages behind authentication or paywalls cannot be accessed
- Very large pages may be truncated to fit the model’s context window