feat: nested eval directories

[allenzhou101]

Add support for nested eval directories. You can now organize evals into folders and use glob patterns to filter them:

evals/
  vercel-cli/
    deploy/
    link/
  flags/
    create/
    update/

Filter examples in experiment config:

• evals: 'vercel-cli/*' - Run all vercel-cli evals
• evals: ['vercel-cli/*', 'flags/*'] - Run multiple categories
• evals: '*/deploy' - Run all deploy evals across folders
• evals: 'vercel-cli/deploy' - Run specific nested eval

Results automatically maintain the hierarchy (e.g., results/experiment/.../vercel-cli/deploy/).

[socket-security]

Review the following changes in direct dependencies. Learn more about Socket for GitHub.

Diff	Package	Supply Chain Security	Vulnerability	Quality	Maintenance	License
	minimatch@​10.1.2

View full report