Python: Document anonymous Parallel Search MCP usage

[George Pickett (georgeatparallel)]

Motivation & Context

The MCP examples show authenticated servers, but lack a runnable example of anonymous public web search and page extraction. This adds one to the existing MCP README using MCPStreamableHTTPTool.

Parallel Fast looks worth trying here. In Artificial Analysis's August 31 Search API results, it scored 80.2/100 on DeepSearchQA F1 and 73.1/100 on the overall Search Index, with 19.2 seconds per task including model and search time. That benchmark tests Search APIs, so it is a reason to try this option rather than a measurement of this MCP example.

The Python harness already enables the selected client's hosted web search by default. Artificial Analysis does not list those hosted implementations. We would need a head-to-head comparison before recommending a new default.

I work at Parallel, which operates this service.

Description & Review Guide

• What are the major changes? A README example with dependency installation, tool discovery, and direct web_search and web_fetch calls through Parallel Search MCP. It reuses one session identifier and closes the connection through the existing context manager.
• What is the impact of these changes? Readers can try both tools without a Parallel key or a model provider account. The guide explains data sent to Parallel, rate limits, and agent activation. Existing providers, defaults, and framework APIs are unchanged.
• What do you want reviewers to focus on? The setup instructions and fit within the MCP guide. The exact snippet passed with prebuilt agent-framework-core 1.17.0 and mcp 1.29.1 in an empty credential environment: both tools were discovered, search returned results, and fetch returned page content without errors. Ruff and the diff whitespace check passed. No repository build or full test suite was run for this documentation-only change.

Related Issue

No linked issue. This small documentation addition uses existing APIs and follows the trivial-change exception in the contributing guide's suggested workflow.

Contribution Checklist

• The code builds clean without any errors or warnings
• All unit tests pass, and I have added new tests where possible
• The PR follows the Contribution Guidelines
• This PR is linked to an issue and there is no other open PR for this issue (see Related Issue above).
• This is not a breaking change. If it is a breaking change, add the breaking change label (or add "[BREAKING]" to the title prefix, before or after any language prefix). A workflow keeps the label and title prefix in sync automatically.

[Copilot]

🟢 Approval recommended

The example matches the MCP API; only a minor entry-point convention issue remains.

Pull request overview

Documents anonymous Parallel Search MCP usage for Python developers.

Changes:

• Adds setup and runnable search/fetch example.
• Documents data sharing, rate limits, cleanup, and agent integration.

File summaries

File	Description
python/samples/02-agents/mcp/README.md	Adds the Parallel Search MCP guide and example.

Review details

• Files reviewed: 1/1 changed files
• Comments generated: 1
• Review effort level: Balanced

---

💡 Add a code-review agent skill for context-aware, tailored reviews. Learn more in the docs.