# Attention Management

> Attention Management

- Skill: `sethdford/attention-management` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds@latest add sethdford/attention-management`
- Raw SKILL.md: https://api.skillmd.com/api/skills/sethdford/attention-management/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: sethdford (https://skillmd.com/u/sethdford)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/sethdford/attention-management

---

# Attention Management

Attention is scarcer than clock time. Design the environment and batch work so depth is possible without heroic willpower.

## When to Use
- Constant context switching, notification fatigue, or shallow days that feel “busy” but empty.
- You want deep work blocks and fewer half-finished threads.

## Workflow
1. **Attention types**: focused (deep work), scattered (shallow), captured (alerts and interruptions)—audit where hours actually go.
2. **Environment first**: remove or distance distractions before the session starts—willpower is a backup, not the plan.
3. **Attention residue**: after switching tasks, the prior task lingers mentally—batch similar work and finish closures before hard switches.
4. **Notification audit**: default off for non-essential channels; batch checks at set times where possible.
5. **Attention contracts**: “For the next 90 minutes, only this task”—write it, start a timer, single tab/app if needed.
6. **For the agent**: if the user is thrashing topics, name the pattern and suggest one protected block or a two-item stack for the next hour.

## Examples
**Example 1:** Morning deep work—phone in another room, Slack paused, one document open, 25/5 or 90-minute box—email only after.

**Example 2:** Research rabbit holes—collect tabs into one list, process in a 30-minute batch, then return to the primary deliverable.