fix(ai-enhancer): guard ruleBasedDescription against empty words (crashes on PascalCase names)

[louisdurden]

Problem

analyze crashes during description generation on any codebase that exports PascalCase or all-caps identifiers:

TypeError: Cannot read properties of undefined (reading 'toUpperCase')
    at ruleBasedDescription (.../ai-enhancer/dist/index.js)
    at applyRuleBasedDescriptions
    at runAnalysis

The crash happens after the full tool surface is computed (backend actions, DB ops, workflows all succeed), so the whole analysis is lost at the last step.

Root cause

In packages/ai-enhancer/src/index.ts, ruleBasedDescription:

const words = name.replace(/([A-Z])/g, ' $1').toLowerCase().split(' ');

When name starts with an uppercase letter (e.g. GET, createPatient, useSoapSign), the regex prepends a leading space, so split(' ') produces a leading empty-string element. The final line then does w[0].toUpperCase() on '', and ''[0] is undefined.

Fix

Trim before splitting and drop empty tokens so words never contains an empty string:

const words = name.replace(/([A-Z])/g, ' $1').trim().toLowerCase().split(/\s+/).filter(Boolean);

Repro

Run analyze against any project with PascalCase exports (a Next.js App Router app reproduces it reliably — hundreds of route.ts / component exports). Before: crash. After: completes and emits descriptions normally.

[vercel]

@louisdurden is attempting to deploy a commit to the Amarnath's projects Team on Vercel.

A member of the Team first needs to authorize it.