Warp's native agent already reads .claude/skills/ -- please extend the same treatment to .claude/commands/.md (custom slash commands) and .claude/agents/.md (subagents).

Today I have to either run Claude Code inside Warp (which means I'm paying $20/month essentially to host another vendor's CLI) or manually port every command to a SKILL.md.

Auto-discovering and registering these as native /commands -- and mapping subagent definitions onto Warp's orchestration/skills primitives -- would let long-time Claude Code users adopt Warp's native agent without rebuilding their workflow. This is the single thing blocking me from switching.