Parallel web search

Parallel Search MCP provides web_search and web_fetch over Streamable HTTP without an API key. Anonymous access is intended for exploration and light use, with rate limits.

The example configuration registers a parallel server and disables editor imports. Passing it explicitly with --config keeps these commands separate from your project and user configuration.

#Run from a checkout

Use Node 24 or 26 and the pnpm version declared in package.json. From the repository root:

pnpm install --frozen-lockfile
pnpm build
node dist/cli.js --config examples/parallel-search/mcporter.json list parallel --schema

Search for sources, then fetch a page for more detail:

node dist/cli.js --config examples/parallel-search/mcporter.json call parallel.web_search \
  --args '{"objective":"Find the official Model Context Protocol documentation","search_queries":["Model Context Protocol official documentation"]}'

node dist/cli.js --config examples/parallel-search/mcporter.json call parallel.web_fetch \
  --args '{"urls":["https://modelcontextprotocol.io/"],"objective":"What is Model Context Protocol?"}'

Search output includes source URLs and excerpts. Fetch output contains page excerpts for the requested URL. Use list parallel --schema to inspect current tool arguments before extending the calls, and add --output json when processing results in a script.

#Use your own configuration

Copy the parallel server entry into your existing mcpServers map to use it alongside other servers. Keep your existing imports setting. The example adds no authorization header or environment placeholder, so no Parallel credentials are needed.

If anonymous requests are rate limited, wait for the server's retry interval before calling again. Higher limits require authenticated access; see the Parallel setup guide.