Fail fast on empty openai-api-key to prevent misleading server-info ENOENT

[topherbuckley]

Summary

Fixes issue #70 by adding explicit preflight validation for openai-api-key and preventing downstream server-info
reads when proxy startup cannot occur.

Root cause

Read server info could run when prompt/prompt-file was set even if openai-api-key was empty. In that case, proxy
startup was skipped, no <run_id>.json was created, and the action later failed with a misleading ENOENT error.

Fix

• Added a preflight step in action.yml (for prompt executions) that trims openai-api-key and fails fast when empty:
  • openai-api-key input is empty. Set OPENAI_API_KEY secret and pass it to this action.
• Updated Read server info step condition to run only when openai-api-key is non-empty.

Before / After

• Before: missing/blank key could fail later with Failed to read server info ... ENOENT.
• After: missing/blank key fails immediately with a clear actionable message; downstream read-server-info does not
  run on invalid startup paths.
• Valid key path: unchanged.

Scope

Minimal, focused change in action.yml only.

[github-actions]

All contributors have signed the CLA ✍️ ✅
_{Posted by the CLA Assistant Lite bot.}

[topherbuckley]

I have read the CLA Document and I hereby sign the CLA