major-ai-skills/skills/set-a-word-count-ceiling/SKILL.md

Installable agentic skills / AI agent skills (SKILL.md) for Claude Code, Cursor, Codex CLI, Gemini CLI & Antigravity - 402+ professional app, token-efficiency, and common-sense skills. SEO/GEO ready.

---

name: set-a-word-count-ceiling
description: "Specify and check a word or sentence limit when a deliverable must fit a constrained format."
category: common-sense
risk: safe
source: self
source_type: self
date_added: "2026-08-26"
tags: ["word-count", "conciseness", "token-savings", "brevity", "information-density", "prompt-engineering"]
tools: ["claude", "cursor", "gemini", "codex", "chatgpt"]
---


# Set a Strict Word Count Ceiling (Token Density) (AI Skill)

## Overview
Because AI models are trained on internet text and RLHF preferences that subtly reward comprehensive answers, they default to writing 300-to-500 word responses for simple questions that only require 40 words.

Setting a **Strict Word Count Ceiling** is one of the highest-leverage constraints in prompt engineering: it forces the AI to prioritize the single highest-value information tokens and delete all conversational padding.

---


## Unconstrained Rambling vs. Bounded Word Ceiling

```
┌─────────────────────────────────────────────────────────────┐
│                 Output Token Density Impact                 │
│                                                             │
│  Unconstrained Prompt ("Explain Docker"):                   │
│  • 450 words generated                                      │
│  • 18 sentences with historical background and marketing    │
│  • High cognitive load, slow reading time                   │
│                                                             │
│  Bounded Ceiling ("Explain Docker in under 40 words"):      │
│  • 32 words generated (92% token reduction)                 │
│  • "Docker packages an application and its dependencies     │
│     into a lightweight container that runs identically on   │
│     any computer or server."                                │
│  • Instant, crystal-clear comprehension                     │
└─────────────────────────────────────────────────────────────┘
```

---


## Channel-Specific Word Ceiling Formulas

Match your word ceiling constraint to your target communication medium:

| Target Channel | Recommended Word Ceiling Constraint | Why |
| :--- | :--- | :--- |
| **Slack / Teams** | *"Keep response strictly under 50 words."* | Fits on mobile screen without a "See more" fold. |
| **Executive Email** | *"Keep response strictly under 100 words."* | Respects executive attention and gets read in 15 seconds. |
| **SMS / Push Alert**| *"Keep response under 25 words."* | Fits in a single SMS segment / push notification. |
| **Landing Page Hero**| *"Under 12 words; no compound sentences."* | Instant visual impact on desktop and mobile browsers. |

---


## Master Word Ceiling Prompt Templates

### Pattern 1: The Hard Ceiling Directive
```markdown
Explain [CONCEPT / QUESTION].

Constraints:
- Word count ceiling: **Maximum 60 words**.
- Deliver the core mechanism in sentence 1.
- State the primary practical use case in sentence 2.
```

---


### Pattern 2: The Sentence Count Guardrail
```markdown
Draft a response to [SITUATION].

Rules:
- Write **EXACTLY 3 sentences**.
- Sentence 1: Acknowledge the problem.
- Sentence 2: State the direct solution and timeline.
- Sentence 3: Call to action / next step.
```

---


## Real-World Comparison

### Scenario: Replying to a Request for a Meeting Reschedule

#### Unconstrained AI Draft (Verbose & Fluffy)
> *"Dear Michael, Thank you for reaching out regarding our scheduled catch-up for tomorrow. I hope your week is going smoothly. I am writing to let you know that an unexpected conflict has arisen on my calendar during our scheduled time slot. I apologize for any inconvenience this may cause you. Would it be possible for us to reschedule our call for later in the week? I am very much looking forward to our discussion..."* *(78 words of repetitive apology).*

#### Constrained Draft (Ceiling: Under 30 Words)
> *"Hi Michael - an unexpected conflict came up for tomorrow. Can we move our sync to Thursday at 2:00 PM EST instead? Let me know if that works!"* *(26 words, clear and respectful).*

---


## Pro-Tip for Enforcing Strict Ceilings
Language models do not count words character-by-character while writing. If you need a strict limit of 100 words, prompt: *"Keep under 80 words"*. Setting the target slightly lower guarantees the final output never exceeds your true upper ceiling.