Add Parallel Search MCP preset

[georgeatparallel]

Summary

• adds Parallel Search to the existing first-class MCP preset catalog, giving users an optional hosted web-search and content-retrieval tool
• uses the native streamableHttp transport at https://search.parallel.ai/mcp, without an API key, account, or authentication requirement
• links setup guidance at the distinct browsable documentation URL: https://docs.parallel.ai/integrations/mcp/search-mcp
• keeps the preset passive and uninstalled until a user explicitly chooses Add; all existing providers, DuckDuckGo behavior, local privacy behavior, and defaults remain unchanged

Data disclosure

No data is sent to Parallel before a user explicitly opts in. Only after the user enables Parallel Search and actively uses its tools are submitted search objectives, search queries, and requested URLs sent to Parallel. This data enables Parallel Search to perform the requested web research and return relevant search results and page content to the agent.

Validation

• isolated Python source assertions for the constructor argument positions, separate docs and server URLs, native transport, post-consent disclosure text, credential absence, and exact one-file scope
• git diff --check
• git status --short

Disclosure: I work at Parallel.ai, which operates this MCP endpoint.