chore: Add integration tests for CLI args to form fields

[richardkmichael]

Add Playwright tests to cover command line argument parsing into form fields.

Motivation and Context

Command line arguments were not populating form fields as expected, which was fixed by #641. Add tests to prevent another regression.

How Has This Been Tested?

Running tests with npm run test:e2e will execute these new tests.

Breaking Changes

None

Types of changes

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to change)
• Documentation update

Checklist

• I have read the MCP Documentation
• My code follows the repository's style guidelines
• New and existing tests pass locally
• I have added appropriate error handling
• I have added or updated documentation as needed

[olaservo]

Thanks for adding these! I think we're good to merge these if you want to remove the extra comments.

[richardkmichael]

This is ready, but I'm uncertain how to proceed, and how much value is here or how much effort
should be put into testing the Inspector.

This PR aims to cover the user-experience -- npx @mcp/i [args] with high testing ROI. (Otherwise,
much separate piece-wise coverage is needed; at least, for parsing of args in both start.js and
cli.ts, several function calls and invocation of subprocess with correct arguments.)

A simple testing approach is used: two scenarios, with different CLI options, result in identical UI
config, hence the same Playwright tests can run in each scenario (i.e., Command input = npm,
etc.).

In this PR as env CLI_TEST_SCENARIO:

  switch (process.env.CLI_TEST_SCENARIO) {
    case "file":
      return "npx . --config client/e2e/fixtures/everything-server.json --server everything";
    case "inline":
      return "npx . -e FOO=bar -e BAZ=bat npm --silent --prefix /path/to/mcp/servers/src/everything run start";
    default:
      return "npm run dev";
  }

controlled from client/package.json:

  "test:e2e:cli-file": "CLI_TEST_SCENARIO=file playwright test --grep @cli e2e",
  "test:e2e:cli-inline": "CLI_TEST_SCENARIO=inline playwright test --grep @cli e2e",

Combinations of new and mutually exclusive CLI options need a new Playwright web server startup command(s). Also, when command line options result in different UI state, then additional test tags would be added to run only the tests for that state.

When more command line options are added (recently, --transport and --server-url), the approach here could be unwieldy. It seems the Inspector is (helpfully) keeping pace with Claude Code for MCP-related options, so I do expect more CLI options will be added here as mcp.json and related config continue developing.

In addition, it looks like the existing CLI tests are not testing CLI parsing, but rather query param handling (via overrides in configUtils.ts), but I could be missing something.

Query param handling tests overlap conceptually with the existing default UI values (i.e., no query params) tests in transport-type-dropdown.spec.ts and should perhaps be moved there. For example, this test is nearly this test (just needs one .expect() to validate the select default).

A few options:

1. Close
2. Merge as is, perhaps has some value
3. Merge after changes:

• Add one --transport ... --server-url ... variant to the inline case
• Move query param parsing to the other spec file (?)

4. Write tests around start.js and cli.ts parsing and runWebClient() to cover a few of the entry pieces.
5. ?

If not 1 or 2, maintainer feedback would be appreciated. :) Is this worthwhile? How much should
Playwright be used; and revisit #582 ?