WEB RESEARCH / Parallel
Parallel Search MCP
Search and fetch web evidence through an anonymous or authenticated hosted endpoint.
“Parallel Search MCP provides web search and fetch tools for MCP-aware clients.”
01 / THE REASONING
Why this made the selection.
- Official hosted integration with documented authentication and account permissions.
- Find source pages for a research question and fetch selected URLs before writing a cited answer.
- Parallel’s page distinguishes a Search MCP that it says is free and does not require an API key from authenticated endpoints that accept OAuth or a bearer API key.
- On the Search endpoint, OAuth is not advertised, so clients that can sign in with OAuth will not prompt on that endpoint.
02 / THE REVIEW RECORD
What we actually inspected.
Source review has boundaries.
A clear record is more useful than a “safe” badge.
Material inspected
- https://docs.parallel.ai/integrations/mcp
- Round 2 reread of acquired publisher setup and permission evidence; see immutable wave review.
Our findings
- The current official guide distinguishes anonymous, bearer and OAuth endpoints and separates Search from Task MCP.
- No public implementation or service redistribution licence is claimed.
- The page says the Search MCP is free to use and that no API key is required for exploration and light use. Higher limits are described as a separate production path.
- OAuth is not advertised on that Search endpoint. Clients that support OAuth sign-in will not prompt for it there.
- A different listed configuration says authentication is required and that the server accepts OAuth and a Parallel API key bearer token.
- The ordinary /mcp endpoint permits anonymous use with optional bearer authentication; /mcp-oauth is a separate endpoint that enforces authentication.
Not established by this review
- Server startup, authentication, live tool execution and security isolation were not tested.
- Published package/container parity with a source revision and hosted implementation internals were not tested.
- The resource was not executed and its tools were not called.
- The server was not called. A source review is not a runtime or security certification.
The review applies to the material and revision named here. A newer upstream release can change its behavior.
03 / PUT IT TO WORK
Connect a server deliberately.
- Add the Streamable HTTP search endpoint for anonymous exploration; no local package or API key is required for that mode.
- For higher limits use a securely supplied API-key bearer header, or select /mcp-oauth for OAuth and mandatory authentication.
- Inspect web_search/web_fetch availability, then keep source URLs and retrieval scope with research outputs.
- Use https://search.parallel.ai/mcp for anonymous exploration, or deliberately choose the OAuth endpoint when organizational authentication is required.
Before you start
- A Streamable HTTP MCP client; a Parallel account only for authenticated usage.
Documented configuration example
Use the client and authentication context described in the setup guidance. Configuration formats vary between hosts.
{
"mcpServers": {
"parallel-search": {
"type": "http",
"url": "https://search.parallel.ai/mcp"
}
}
} 04 / FOLLOW THE EVIDENCE
The source trail.
Our notes are separate from the original resource.
Check upstream before adopting a new version.
https://docs.parallel.ai/integrations/mcp
Supports: summary, install, configExample, access, limitations, bestFor, review, compatibility, transports