feat(skills): add session-quality-gate — session-end learning verification

[YuhaoLin2005]

What this is

A session-end verification skill — asks "did this session produce transferable learning, or did we just ship and forget?" before allowing work to conclude.

Four-step process: Self-Audit → Learning Capture → Disk Check → Rationalization Detection. Works with any project structure — adapt the capture paths to whatever you use (Notion, Obsidian, plain Markdown, nothing at all).

No specific file system required. The verification framework is the value; wire it to your own capture system.

How it complements this repo

agent-skills has production-readiness and code-quality covered. It doesn't have session-hygiene — the discipline of verifying that work finished clean before moving on.

Existing skill	Checks	This adds
shipping-and-launch	Deploy safety	Thinking completeness at session end
code-review-and-quality	Code correctness	Learning capture quality
doubt-driven-development	Surface uncertainty during work	Audit completeness after work

Same verification-gate architecture (checklist + rationalizations + red flags), different dimension — thinking quality rather than code or deploy quality.

Structure

• Self-Audit → Learning Capture → Disk Check → Rationalization Detection
• Common rationalizations with rebuttals
• Disk thresholds with concrete references

Extracted from practice

Derived from a learning-capture workflow used across 200+ sessions. Before verification, same debugging mistakes repeated. After, cross-session knowledge transfer became measurable.