Clear the CLI startup timeout when we are told to stop

[igfoo]

With index.ts as

import { CopilotClient } from "@github/copilot-sdk";

const client = new CopilotClient();
await client.start();
await client.stop();

running npx tsx index.ts takes more than 10 seconds. This is problematic for scripts that run quick queries, or even decide to run no queries after starting a client.

This is due to the 10 second timeout that is used to declare the CLI server failed to start.

With this PR, the above takes less than a second.

I can't run the npm test tests locally on main; I get as far as

 Test Files 3 passed (30)
      Tests 72 passed (72)
   Start at 17:36:49
   Duration 624.16s

and then it seemingly makes no further progress. But I get the same on my branch, and I assume that CI has run them.

[Copilot]

Pull request overview

This PR addresses a Node.js SDK shutdown delay by tracking the “CLI startup timeout” timer and clearing it when the client is stopped, so short-lived scripts don’t remain alive waiting for the 10s startup timeout to elapse.

Changes:

• Track the CLI startup timeout handle on CopilotClient (cliStartTimeout).
• Clear the pending startup timeout during stop() and forceStop().
• Store the timeout handle when starting the CLI server.

Show a summary per file

File	Description
nodejs/src/client.ts	Track and clear the CLI startup timeout so stop/forceStop can exit promptly instead of being held by a pending 10s timer.

Copilot's findings

Comments suppressed due to low confidence (2)

nodejs/src/client.ts:1538

• cliStartTimeout is set here but never cleared when the CLI successfully starts or when startup fails (e.g., error/exit handlers). If start() throws and the caller does not invoke stop(), this pending timeout can keep the Node.js event loop alive for up to 10 seconds. Consider clearing+nulling this.cliStartTimeout on all completion paths (resolve/reject), and also clearing any existing timeout before assigning a new one to avoid orphaned timers on concurrent start() calls.

            // Timeout after 10 seconds
            this.cliStartTimeout = setTimeout(() => {
                if (!resolved) {
                    resolved = true;
                    reject(new Error("Timeout waiting for CLI server to start"));

nodejs/src/client.ts:624

• The clearTimeout + nulling logic is duplicated in both stop() and forceStop(). To reduce the chance of the two paths diverging over time, consider extracting this into a small private helper (e.g., clearCliStartTimeout()) and calling it from both places.

        if (this.cliStartTimeout) {
            clearTimeout(this.cliStartTimeout);
            this.cliStartTimeout = null;
        }

• Files reviewed: 1/1 changed files
• Comments generated: 1

[Copilot]

There’s now important cleanup logic for cliStartTimeout, but there are currently no unit tests covering stop()/forceStop() behavior (this file’s tests mainly use forceStop() as cleanup). Adding a small unit test with fake timers to assert that stop() (and forceStop()) clears a pending startup timeout would help prevent regressions where short-lived scripts hang for ~10s again.

This issue also appears in the following locations of the same file:

• line 621
• line 1534