[Feature]: Add an opt-in Parallel Search MCP research-to-memory example #485

Area

use cases

Problem or opportunity

EverOS can retain research across sessions, but this repository has no runnable example connecting web search and page fetching to its memory lifecycle. Users currently have to assemble that workflow themselves.

Proposed solution

Would an opt-in research-to-memory example fit under examples/? Parallel Search MCP provides free web search in Fast mode and page fetching, with no account or API key required.

I would add a small Python example using Streamable HTTP at https://search.parallel.ai/mcp. It would search a topic, fetch a selected source, retain URLs with the excerpts, then use EverOS's existing /api/v2/memory/add, flush, and keyword-search flow to store and recall the research. Dependencies and setup would stay local to the example, preserving runtime defaults and saved configuration.

The example would require a running EverOS server with its extraction model configured. I would cover search/fetch failures and validate the full search-to-memory-to-recall path.

Alternatives considered

Users can configure search and memory separately in an MCP-capable assistant. A runnable example would make the handoff and source provenance explicit without adding web-search dependencies to EverOS core.

Additional context

Following the feature/use-case proposal route in CONTRIBUTING.md.

I work at Parallel.