Add more detection trigger patterns

The current trigger patterns cover common cases but there are more natural ways people express preferences:

**Potential high-confidence patterns to add:**
- "Stop doing X"
- "Please always X"
- "I need you to X"
- "Rule: X"
- "Important: X"

**Potential medium-confidence patterns to add:**
- "I tend to X"
- "I usually X"
- "My convention is X"
- "In this project we X"

**To contribute:** Edit `skills/active-listening/SKILL.md` and add patterns to the appropriate confidence section. Add corresponding test cases in `tests.json`.

Start with skills/active-listening/SKILL.md and inspect the existing high- and medium-confidence trigger sections. Add the listed patterns in the appropriate sections, then add corresponding cases to tests.json; done means each proposed pattern is represented at the intended confidence level and covered by the test data.

Written by the indexing model from the issue text.