feat(sheets): add sheets.appendRow write tool

[rohanaceres]

Summary

• Adds sheets.appendRow MCP tool to append rows to a Google Sheets spreadsheet
• Tool is gated behind the sheets write feature group (defaultEnabled: false, requires spreadsheets OAuth scope)
• Uses USER_ENTERED value input option so formulas and dates are parsed correctly

Test Plan

• Unit tests added for success case, URL-to-ID extraction, and error handling (SheetsService.test.ts)
• Run npm run test && npm run lint to verify all checks pass
• Manually verify appending a row to a spreadsheet returns correct updatedRange, updatedRows, updatedColumns, updatedCells
• Verify tool is disabled by default and only activates when sheets write feature is enabled

Issue #341

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.

[gemini-code-assist]

Code Review

This pull request introduces a new appendRow tool for Google Sheets, enabling the addition of multiple rows to a spreadsheet. The implementation includes the core service logic, tool registration with Zod validation, and unit tests. Review feedback suggests renaming the tool and its associated methods to appendRows to accurately reflect its multi-row capability and recommends broadening the input schema and TypeScript types to support numbers, booleans, and null values in addition to strings.

[allenhutchison]

Review summary

Clean, idiomatic addition — appendRows mirrors the sibling getRange method exactly (ID extraction via extractDocId, logToFile start/finish/error, the try/catch result shape), the feature gating is correct (sheets/write, defaultEnabled: false, spreadsheets scope), and the happy-path test asserts the full request shape including valueInputOption: 'USER_ENTERED'. Nothing here is merge-blocking. I've left inline notes on a phantom-success edge case, the isError convention, and an empty-values guard; grouping the docs and test items here.

Docs parity (please update before merge)

The repo enumerates every tool per feature group, but the new write tool isn't listed:

• docs/index.md (Google Sheets section, ~line 42): add a bullet, e.g. - `sheets.appendRows`: Appends rows to a Google Sheets spreadsheet.
• docs/feature-configuration.md: there's a ### sheets.read section (line 194) but no ### sheets.write section — every other write group (docs.write, drive.write, gmail.write, …) has one. The scope table (line 25) also lists only sheets | read with no sheets | write row. Both need a sheets.write entry listing sheets.appendRows.

Test coverage

The new tests cover the happy path (with full request-shape assertion), URL-to-ID extraction, and the thrown-error path — good, and consistent with the existing getRange/getMetadata style. Two gaps worth closing:

• response.data.updates === undefined — the highest-value addition, directly exercising the phantom-success path flagged inline. Mock { data: {} } and assert the result reflects "no rows written."
• Empty values: [] — documents whether an empty append is rejected locally (it would be, if .min(1) is added) or passed through.
• If isError: true is added, a regression test asserting result.isError === true on the error path would lock it in (see DriveService.test.ts / DocsService.comments.test.ts for the pattern).

Minor

• The PR title/body say sheets.appendRow (singular), but the tool is registered as sheets.appendRows (plural) everywhere in code — the code is internally consistent; just the description is off.
• The URL-extraction test uses an unrealistically short ID (abc123); a real-style ID like 1A2b-_C3d would also guard the -/_ character class in the ID regex.

Strengths

• Faithful to the established getRange pattern, so it reads as a natural sibling.
• Correct off-by-default gating; the existing feature-config test still passes.
• USER_ENTERED matches the stated intent, and the string | number | boolean | null cell union is exactly right (allows null, excludes undefined).

Review assisted by automated analysis agents; findings verified against the diff.

[allenhutchison]

The four fields are read with response.data.updates?.… optional chaining. If updates is undefined — which the API can return on a 200 where nothing was written — JSON.stringify drops all four undefined fields and the tool returns {}, plus the "Finished appendRows" log line, which is indistinguishable from a real success. For a write tool, a phantom success can mean silently lost data with no signal to the caller. Suggest treating a missing updates as a failure rather than masking it:

const updates = response.data.updates;
if (!updates) {
  logToFile(`[SheetsService] appendRows returned no update metadata for: ${id}`);
  return {
    isError: true,
    content: [{ type: 'text' as const, text: JSON.stringify({
      error: 'Append returned no update information; rows may not have been written.',
    }) }],
  };
}
// then read updates.updatedRange, etc. directly (no `?.`)

[allenhutchison]

This error result isn't flagged with isError: true, so the MCP client treats the call as a success and has to string-parse the JSON body to notice the error key. DriveService (handleError) and DocsService already set isError: true on failures. Note this is consistent with the other Sheets methods (getText/getRange/getMetadata all omit it), so it's a pre-existing pattern rather than something this PR introduces — but since appendRows is a write operation where a false "success" is more costly, it's worth adding here (ideally to all four in one pass):

return {
  isError: true,
  content: [{ type: 'text' as const, text: JSON.stringify({ error: errorMessage }) }],
};

[allenhutchison]

values: [] currently passes validation and forwards an empty requestBody.values to the API — a wasted no-op call. A .min(1) guard rejects it at the tool boundary with a clear message: z.array(z.array(z.union([...]))).min(1). (Don't enforce rectangular rows — ragged rows are legal in Sheets, where shorter rows just leave trailing cells untouched.)