← Back to videos

3 Claude Code Features Most People Miss: Sub-Agents, /status, and Auto-Skill Hooks

@qendresahhotiCapturedClaude Code / Skills
See more in Claude Code / Skills

Exact instruction

  1. Sub-agents: Type 'spin up a sub agent to research [topic]' — Claude creates a separate instance, does the work, returns only the summary.
  2. Context monitoring: Type /status for color-coded context meter; /context for breakdown by component.
  3. Auto-skill hooks: Write a hook script that reads your prompt and loads the matching skill automatically. Follow @qendresahhoti for setup details.

Suggested prompt

When starting a research task within an existing session: 'Spin up a sub agent to research [topic] and return just the key findings as a summary. Keep the full research in the sub-agent context, not here.'

Adopt?

Yes: All three are immediately actionable. Sub-agents solve context degradation on long sessions. /status + /context enable proactive context management before sessions break. Auto-skill hooks eliminate the overhead of manually specifying skills every time.

Find the resource

Follow @qendresahhoti for setup details on the auto-skill hook script. The /status and /context commands are built into Claude Code — no setup needed.

show original caption

Three things inside Claude Code that most people don't know exist.Sub-agents: instead of doing everything in your main session and burning through context, spin up a separate instance that does the heavy lifting and sends back just the summary. Your main session stays clean.Context management: type /status line to see a color coded meter of how ful