# Just Do It Manually Skills

> Just Do It Manually

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

---

# Just Do It Manually

> You've spent 30 minutes designing automation for a 5-minute task. Stop. Just do the thing.

---

## When to Use

- You're building a system to avoid doing the work
- You're "optimizing" something you've done once
- The automation is more interesting than the task
- You've spent more time on the meta than the actual

**Trigger:** Spent >15 minutes designing/automating a task that takes <10 minutes to do manually.

---

## The Anti-Productivity Skill

This is a contrarian skill. Most productivity advice says "automate everything."

**Reality check:** Sometimes automation is procrastination in disguise.

You're not being efficient. You're:
- Avoiding the boring work
- Enjoying the meta-work more than the work
- Perfectionism disguised as optimization
- Building a system you'll use once

---

## The Smell Test

### Red Flags (Just Do It Manually)

- ⚠️ You've never done this task before
- ⚠️ You'll do it fewer than 5 times total
- ⚠️ The automation is "interesting" and the task is "boring"
- ⚠️ You keep adding features to the automation
- ⚠️ You're avoiding starting the actual work
- ⚠️ The task itself would take <10 minutes

### Green Flags (Automation Makes Sense)

- ✅ You do this task weekly or more
- ✅ The task is error-prone when done manually
- ✅ Others will use the automation too
- ✅ The automation is simpler than the manual process
- ✅ You've done it manually several times already

---

## The Protocol

### Step 1: Time Check (10 seconds)

> "How long would this task take to do manually, right now?"

Be honest. Not the worst case. The realistic case.

### Step 2: Automation Check (10 seconds)

> "How long have you already spent (or will you spend) automating this?"

Include:
- Research time
- Building time
- Testing time
- Documentation time

### Step 3: The Math (5 seconds)

> "Automation time ÷ Manual time = Break-even reps"

**Example:**
- Manual: 5 minutes
- Automation: 60 minutes
- Break-even: 12 times

> "Will you do this task 12+ times? If not, just do it manually."

### Step 4: The Honesty Check (10 seconds)

> "Are you automating because it's useful, or because it's more interesting than the actual work?"

If it's the latter: **stop, do the thing, move on.**

---

## XKCD Reference

The famous [XKCD chart](https://xkcd.com/1205/):

| Task frequency | Shave 1 min | Shave 5 min | Shave 30 min |
|----------------|-------------|-------------|--------------|
| Daily | 6 hours/5yr | 1 day/5yr | 1 week/5yr |
| Weekly | 1 hour/5yr | 4 hours/5yr | 1 day/5yr |
| Monthly | 12 min/5yr | 1 hour/5yr | 4 hours/5yr |
| Yearly | 1 min/5yr | 5 min/5yr | 30 min/5yr |

**Most things we automate are monthly or less.** The payoff is tiny.

---

## The Manual-First Rule

Before automating anything:

1. **Do it manually 3 times first**
   - You'll understand the edge cases
   - You might find you don't need to do it again
   - The automation will be better informed

2. **Write down the steps as you go**
   - This IS the documentation
   - If you need to automate later, you have a spec

3. **Only automate if you hit the 4th time**
   - Now you know it's recurring
   - Now you know the actual process

---

## Common Over-Engineering Traps

### The "just in case" system
> Building for flexibility you'll never use. YAGNI (You Ain't Gonna Need It).

### The rabbit hole
> "While I'm at it, I might as well add..." No. Ship the minimum.

### The sunk cost trap
> "I've already spent 2 hours, might as well finish." No. Cut losses.

### The perfectionism mask
> "I want to do it right." Translation: "I don't want to do it at all."

---

## The Decision

After running this check, choose:

**Option A: Just do it manually**
> Close the automation. Do the task. It'll take 10 minutes. Then it's done.

**Option B: Continue automating (with constraints)**
> Set a timebox: "I have 30 more minutes. If it's not done, I do it manually."

**Option C: Delegate it**
> If you don't want to do it AND you don't want to automate it, maybe it shouldn't be done at all. Or pay someone.

---

## Output

By end of check:
1. Honest assessment: Am I automating or avoiding?
2. The math: Does automation break even?
3. A decision: Manual / Timebox / Drop

---

*Sometimes the most productive thing is to stop being "productive" and just do the work.*