Add browser-native Parallel Search tools for WebMCP websites - #51
Draft
georgeatparallel wants to merge 7 commits into
Draft
Add browser-native Parallel Search tools for WebMCP websites#51georgeatparallel wants to merge 7 commits into
georgeatparallel wants to merge 7 commits into
Conversation
Contributor
Author
|
Tracking this in DEV-259, assigned to me. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Agents can only use tools they know about. Today, an agent can be browsing a website and still have no way to search the rest of the web unless someone already set up a search provider for it. That makes distribution depend on getting each agent product to integrate us first.
WebMCP opens up a different path: the website itself can offer tools to compatible agents while they are on the page. This PR makes Parallel one of those tools. A site owner adds one npm package or script tag, and visiting agents get
parallel_web_searchandparallel_web_fetch. Both call our existing free Search MCP, so there is no API key to expose, no backend to build, and no new service to run.The important part for Parallel is that every participating website becomes a distribution point. Instead of waiting for every agent vendor to choose us, a website can make Parallel search available to whatever supported agent shows up. It only works on sites that opt in, and it does not replace remote MCP.
We kept the implementation small: one browser module, two read-only tools, anonymous per-tab sessions, bounded untrusted results, cancellation, and an optional self-installing script. It handles Unicode, failed fetches, and free-tier limits. The browser bundle is about 4.3 KB.
Requests send the search terms or requested URL and an anonymous session ID to Parallel. The referrer includes only the site's origin, not its path or query string. Fetch URL fragments and browser credentials are not sent, and page content and agent history are not collected automatically. Cross-origin tool access requires explicit opt-in.
Validation includes 23 focused tests and 163 total workspace and release tests. All six CI checks pass.
One release detail: merging does not publish a new package automatically. The publishing guide now includes the exact first-release commands an npm organization owner needs to publish it and configure trusted publishing.