fix: add untrusted data wrapper around logs response MCP-197

[nirinchev]

Proposed changes

Part of MCP-197 - this adds the untrusted data wrapper around the logs response and updates the tests to reflect the new format.

[Copilot]

Pull Request Overview

This PR updates the MongoDB logs tool to wrap log response data in an untrusted data wrapper as part of security improvements (MCP-197). The change ensures that log data retrieved from MongoDB is properly marked as untrusted content when returned to users.

• Replaced direct content array with formatUntrustedData wrapper for log responses
• Updated test assertions to handle the new untrusted data format
• Added helper function import for extracting data from untrusted content

Reviewed Changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 2 comments.

File	Description
src/tools/mongodb/metadata/logs.ts	Wraps log data in untrusted data format and simplifies response structure
tests/integration/tools/mongodb/metadata/logs.test.ts	Updates test expectations to handle untrusted data wrapper format

---

_{Tip: Customize your code reviews with copilot-instructions.md. Create the file or learn how to get started.}

[Copilot]

The split('\n') operation may produce empty strings at the end if the logs data ends with newlines. This could cause the length checks and JSON parsing to fail. Consider filtering out empty strings: getDataFromUntrustedContent(elements[1]?.text ?? "").split("\n").filter(line => line.trim())

Suggested change

	const logs = getDataFromUntrustedContent(elements[1]?.text ?? "").split("\n");
	const logs = getDataFromUntrustedContent(elements[1]?.text ?? "")
	.split("\n")
	.filter(line => line.trim());

[Copilot]

Same issue as above - the split('\n') operation may produce empty strings that could break JSON parsing. Consider filtering out empty strings: getDataFromUntrustedContent(elements[1]?.text ?? "").split("\n").filter(line => line.trim())

Suggested change

	const logs = getDataFromUntrustedContent(elements[1]?.text ?? "").split("\n");
	const logs = getDataFromUntrustedContent(elements[1]?.text ?? "")
	.split("\n")
	.filter(line => line.trim());

[himanshusinghs]

Looks good. Do you have something in mind for export tool? I think we can wrap the data there but we need to check if LLM is then able to follow up on that data for further work.

[nirinchev]

Haven't started on the export yet, will check if the LLM gets confused by the wrapper.