Parallel MCP Servers

About

Parallel MCP integrations let AI assistants use Parallel web search and research task capabilities without writing a custom API integration.

Key Features

• Connect an assistant to Parallel web research tools.
• Choose the Search or Task integration appropriate to the workflow.

Use Cases

Find current web information and retrieve sources for an agent research task.

Limitations

The Search and Task integrations serve different workflows; use the endpoint documented for the selected integration.

Sources

• Project documentation: https://docs.parallel.ai/integrations/mcp/quickstart
• Setup reference: https://docs.parallel.ai/integrations/mcp/task-mcp

MCP Setup

Access requirements

• Search: anonymous access with limits; optional API key for higher limits.
• Task: Parallel authentication required.

Client setup

Other

Transport: http

• Add the remote MCP URL: https://search.parallel.ai/mcp

Search endpoint; no OAuth prompt on this URL. Use /mcp-oauth for the documented OAuth variant.

Claude Code

Transport: http

claude mcp add --transport http Parallel-Task-MCP https://task-mcp.parallel.ai/mcp

Auth:

• Run /mcp and complete Parallel sign-in in the browser.

Codex

Transport: http

codex mcp add parallel-task --url https://task-mcp.parallel.ai/mcp --bearer-token-env-var PARALLEL_API_KEY

• Set PARALLEL_API_KEY before launching Codex; restart after adding the server.

Environment variables

• PARALLEL_API_KEY (Task API-key configuration)

Setup docs: https://docs.parallel.ai/integrations/mcp/task-mcp