Goose remote extension web search guide
How to Add Web Search to Goose with Webstractor MCP
Goose calls MCP integrations extensions and supports hosted Streamable HTTP servers directly. Webstractor can provide focused search without installing a package or attaching browser automation, which is useful for research tasks where ranked public sources are the desired output.
Quick answer
Run Webstractor as a remote Goose extension.
For a one-off session, start Goose with --with-streamable-http-extension https://webstractor.com/mcp. For a persistent setup, run goose configure, choose Add Extension, select Remote Extension (Streamable HTTP), and enter the same endpoint.
goose session --with-streamable-http-extension "https://webstractor.com/mcp"Before you begin
What you need
- Goose Desktop or CLI
- A configured model/provider
- Network access to the hosted endpoint
Step-by-step setup
Connect Webstractor to Goose
Try a one-session connection first
Launch Goose with the remote extension flag. This validates transport and tools without making a persistent configuration change.
goose session --with-streamable-http-extension "https://webstractor.com/mcp"Add a persistent remote extension
Run goose configure, select Add Extension, then Remote Extension (Streamable HTTP). Name it Webstractor and enter the endpoint URI.
Set a practical timeout
Keep the default suggested timeout for initial testing. No custom headers are needed for anonymous allowance.
Confirm the extension is enabled
Open Extensions in Goose Desktop or inspect the configured extension in CLI. Verify search_web is among its tools.
Run a provenance-focused query
Prompt: “Use Webstractor search_web to find five official sources about Goose remote extensions. Summarize the configuration choices and include each source URL.”
Extension choice
Focused search is different from computer control
Goose includes extensions that can interact with a browser or desktop. Use those when the task requires clicking and visual interaction. Use search_web when the desired artifact is an ordered list of public sources with compact metadata.
A focused tool generally needs fewer actions and produces a cleaner audit trail for research.
Session design
Use one-off flags for experiments
The --with-streamable-http-extension flag is ideal for testing a tool or creating a reproducible command. Move the server into persistent configuration once it becomes part of daily workflows.
- One task: session flag
- Daily use: goose configure
- Anonymous use: no custom header
- Account usage: add a Bearer header through the remote extension prompt
Troubleshooting
Select Remote Extension, not Command-line Extension
Webstractor is hosted. Choosing Command-line Extension will prompt for a local executable that does not exist. Select Remote Extension (Streamable HTTP) and paste the HTTPS endpoint.
If Goose chooses another search extension, disable the overlap temporarily or explicitly name Webstractor search_web.
Available data
What you can extract
- Ordered public-web results
- Titles, snippets, and traceable URLs
- Site, language, and country controls
- Related tools for news, images, videos, places, and public pages
AI workflows
Where normalized data helps
- Research an unfamiliar repository dependency
- Collect sources before drafting technical documentation
- Find official setup guides for an automation task
- Discover public pages without controlling a browser
Boundaries
Public data only
- Goose has other web-capable extensions; name Webstractor during validation.
- A remote extension requires network access even when the model is local.
- Search results remain untrusted public data.
webstractor.com does not bypass CAPTCHAs, login walls, paywalls, access controls, or regional restrictions. Review the source’s terms and applicable law before collecting or reusing data.
Common questions
Goose and Webstractor FAQ
Do I need npx or Python for Webstractor in Goose?
No. Goose connects directly to the hosted Streamable HTTP endpoint.
Is the CLI flag persistent?
No. It applies to that session. Use goose configure to add a persistent remote extension.
How do I use account credits?
Configure a custom Authorization header with Bearer ext_live_… when the Goose remote extension setup prompts for headers. Keep the value out of shared files.
Ready to try it?