chore(verify): normalize line endings in SKILL.md and auth.md

[bensonwong]

Summary

• Normalize CRLF → LF line endings throughout skills/verify/SKILL.md
• Remove trailing newline from skills/verify/rules/auth.md
• No content changes — whitespace/encoding cleanup only

Test plan

• Confirm diff shows no semantic content changes (line endings only)
• Verify /verify skill still loads and executes correctly

[claude]

Code Review

Summary: Whitespace-only cleanup -- CRLF to LF normalization across SKILL.md (214 lines) and rules/auth.md (15 lines), plus removal of the trailing newline from auth.md. I verified the diff manually: no semantic content was changed.

Code quality / correctness

No issues. The content of both files is identical before and after; only the line-ending bytes differ. The trailing-newline removal from auth.md is harmless and consistent with LF style.

Potential bugs

None introduced. This is a safe, non-functional change.

Performance / Security

N/A.

Test coverage

The PR test plan mentions manually confirming the /verify skill still loads and executes correctly. Since no logic changed, a smoke test of the skill is sufficient.

---

Suggestions

1. Prevent regression with .gitattributes

Without a .gitattributes file enforcing LF, any Windows contributor or editor configured for CRLF can re-introduce mixed line endings. Consider adding one at the repo root:

* text=auto eol=lf
*.md text eol=lf

This makes the fix permanent rather than a one-time cleanup.

2. auth.md trailing newline (minor nit)

Most POSIX tools and editors expect a final newline in text files. If the removal is intentional style, that is fine; otherwise, one trailing newline is the convention.

---

Verdict

Looks good to merge. The change is safe, narrow in scope, and improves consistency. The only follow-up worth considering is adding .gitattributes to enforce the LF policy going forward.