GeezerVison

This PR hardens proxy startup reliability by converting a silent false-success scenario into an explicit failure.

Fixes #3530 issue

Previously, the startup loop could exhaust retry attempts without throwing a terminal error. In such cases, outputs could still be set and execution would continue even though the proxy process was not running.

Before:
Proxy started successfully
[process exited immediately]

After:
Failed to start proxy after 5 attempts (last exit code: 1)

This change ensures that proxy initialization fails fast if stabilization does not occur.

Root Cause

The previous control flow:

• Retried startup attempts
• Only threw on immediate spawn error
• Did not explicitly fail after retry exhaustion
• Could report success if no process remained alive

This allowed workflows to proceed with a non-functional proxy.

Changes

1. Extracted startup logic

Created src/start-proxy/launcher.ts:

• Encapsulates startup behavior.
• Allows dependency injection for testing.
• Improves separation of concerns.

2. Explicit retry exhaustion handling

• Tracks lastExitCode.
• Verifies process remains alive after stabilization delay.
• Throws Error("Failed to start proxy after 5 attempts (last exit code: X)") when no stable process remains.

3. Tests

Added launcher.test.ts covering:

• Retry exhaustion with exit code diagnostics.
• Spawn error propagation.
• Successful stabilization and output configuration.

Backward Compatibility

• No change to output names
• No change to telemetry format
• No change to wrapper success/failure reporting
• No new inputs introduced

Behavioral impact is limited to converting silent false-positive startup into explicit failure.

Testing

Executed:

• npm run lint
• npm run build
• npm run test

Added tests in:
src/start-proxy/launcher.test.ts

All new launcher tests pass on Node 24+.

Covering:

• Exhausted retries
• Spawn error propagation
• Stable process success path

Why This Matters

This improves workflow reliability by:

• Ensuring failures surface at the correct lifecycle stage
• Reducing downstream noise
• Providing clearer diagnostics

Fail-fast behavior improves user debugging experience and system robustness.