[BUG] skill-creator: run_loop.py description improvement step requires ANTHROPIC_API_KEY, incompatible with Claude Code auth

Preflight Checklist

• [x] I have searched existing issues and this hasn't been reported yet
• [x] This is a single bug report (please file separate reports for different bugs)
• [x] I am using the latest version of Claude Code

What's Wrong?

run_loop.py uses the Anthropic Python SDK directly to call Claude for the "improve description" step. This requires ANTHROPIC_API_KEY to be set. Claude Code uses its own authentication mechanism that is not exposed as an environment variable, so this step fails for all Claude Code users who don't separately manage an API key.

What Should Happen?

Use claude -p (already used in run_eval.py for trigger evaluation) for the description improvement step as well, so it works with Claude Code's native auth.

Error Messages/Logs

Steps to Reproduce

1. Run skill-creator on any machine where Claude Code is the only Anthropic client configured (no ANTHROPIC_API_KEY in environment)
2. Run run_loop.py for description optimization
3. Trigger evaluation passes (uses claude -p)
4. Description improvement step fails: TypeError: Could not resolve authentication method. Expected either api_key or auth_token to be set.

Claude Model

Sonnet (default)

Is this a regression?

Yes, this worked in a previous version

Last Working Version

_No response_

Claude Code Version

2.1.92

Platform

Anthropic API

Operating System

Windows

Terminal/Shell

Windows Terminal

Additional Information

Windows 11 Pro 10.0.26200