fix(skills): drop $ARGUMENTS from detail-bugs and detail-create-rules

[sachiniyer]

Summary

• Both skills currently template $ARGUMENTS into commands like detail bugs list $ARGUMENTS. When the skill is invoked with prose ("triage today's bugs"), Claude Code forwards the full sentence as $ARGUMENTS, producing detail bugs list "Check for detail bugs..." — the CLI treats that as a literal owner/repo and every command in the skill fails.
• The CLI already infers the repo from the git remote, and Claude can pass an explicit flag if a different repo is needed, so the templating doesn't carry its weight.
• This drop-the-arg approach is the cleaner alternative to the shape-checking half-fix in usedetail/detail#11222.

Changes

• Removed argument-hint: <owner/repo> from both skill frontmatters.
• Removed every $ARGUMENTS interpolation from command examples.
• Replaced the "Determining the Repository" section in each skill with a single sentence pointing at git-remote inference and explicit flags.

Test plan

• grep -r '\$ARGUMENTS' .claude/skills/ returns nothing
• grep -r 'argument-hint' .claude/skills/ returns nothing
• Invoke detail-bugs skill with a prose request and confirm the detail bugs list command runs cleanly
• Invoke detail-create-rules skill and confirm the detail rules create / detail rules propose commands run cleanly

🤖 Generated with Claude Code