Slash command expansion has no visual indicator to the user

Problem

When a slash command (skill) is expanded by Claude Code, the user receives no visual indicator that the skill was loaded. The skill content is injected into the model's context via <command-name> / <command-message> tags, but these are invisible in the UI. The user cannot distinguish between "Claude is following a loaded skill" vs "Claude is responding on its own."

Steps to Reproduce

Case 1 — Standalone slash command (expands, but no indicator):

1. Create a skill at ~/.claude/skills/my-skill/SKILL.md
2. Type /my-skill as a standalone message
3. Claude follows the skill instructions correctly
4. User sees no indication that a skill was loaded — response looks identical to a normal reply

Case 2 — Embedded in text (does NOT expand, also no indicator):

1. Same skill as above
2. Type "run /my-skill please" (embedded in natural text)
3. Slash command is NOT expanded — Claude receives plain text only
4. User also sees no indication — both cases look identical from the user's side

This is especially confusing with disable-model-invocation: true skills, where Case 2 silently fails (no expansion, model can't invoke via Skill tool either) but the user has no way to know.

Expected Behavior

Some visual indicator when a skill is expanded, e.g.:

• A status line: Using skill: /track
• A badge or icon next to the response
• The expanded skill content shown in a collapsible block
• A distinct message bubble showing the command was recognized

Why This Matters

• Can't tell if the command triggered — user has no feedback loop
• Hard to debug skill failures — when a skill doesn't expand (Case 2), the user doesn't know why the behavior differs
• Reduces trust in the skill system — "did it actually use my skill or just wing it?"

Environment

• Claude Code CLI
• Skills with disable-model-invocation: true in ~/.claude/skills/