# No Slop Grenade

> Use when drafting Slack messages, chat replies, emails, PR comments, code review responses, or any synchronous/conversational reply where the user asked a direct question. Triggers on requests to "write a Slack message", "reply to this", "respond to", "draft a message", or when about to paste long AI-generated text into a chat-style medium. Prevents dumping massive AI-generated walls of text where a human would write one or two sentences.

- Skill: `ebellefontaine/no-slop-grenade` (Agent Skill)
- Install (CLI): `npx skillmds@latest add ebellefontaine/no-slop-grenade`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ebellefontaine/no-slop-grenade/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Integrations & APIs
- Author: ebellefontaine (https://skillmd.com/u/ebellefontaine)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/ebellefontaine/no-slop-grenade

---


# No Slop Grenade

## Overview

A "slop grenade" is pasting a massive AI-generated response into a conversation where a human would write one sentence. Use AI to make things **clearer, not longer**. Sharpen judgment; do not substitute for it.

Source: https://noslopgrenade.com/ (related: https://nohello.net/)

## Core Principle

**When someone asks you (a human) a question in a conversational medium, they want your judgment in one or two sentences -- not a ChatGPT essay.**

Violating the form of conversation violates the intent of conversation.

## When To Apply

Apply on any of:

- Drafting Slack, Teams, Discord, IM, or chat replies
- Composing short emails or replying inline
- Writing PR review comments or issue replies
- Responding to a direct question from a teammate
- Any synchronous or near-synchronous human-to-human medium

Do NOT apply to:

- Formal docs, RFCs, design specs, runbooks (long form is correct there)
- Code, commit messages, or structured artifacts requested explicitly
- Reports or analyses the user explicitly asked to be comprehensive

## Rules

1. Match the medium. Chat gets chat-length replies; docs get doc-length content.
2. Lead with the answer in the first sentence.
3. Cut preamble ("Great question!", "Certainly!", "I'd be happy to...").
4. Cut postamble ("Let me know if...", "Hope this helps!", restating the question).
5. No headers, no bullet lists, no tables unless the recipient asked for structure.
6. Strip generic caveats and AI hedging ("It depends on your context", "Generally speaking").
7. If unsure between short and long: choose short. Offer to expand.
8. Never paste AI output verbatim into a chat. Read it, compress to your voice.

## Quick Reference

| Situation                              | Target length         |
| -------------------------------------- | --------------------- |
| Slack/IM direct question               | 1-3 sentences         |
| Slack thread reply                     | 1-5 sentences         |
| Short email reply                      | 2-5 sentences         |
| PR comment / code review nit           | 1-2 sentences         |
| PR comment / substantive review        | 1 short paragraph     |
| Design doc, RFC, runbook (NOT a chat)  | Long form as needed   |

## Before/After

Question in Slack: "Should we use Redis or Memcached for the session cache?"

```
BAD (slop grenade):
Great question! When deciding between Redis and Memcached for session
caching, there are several factors to consider:

1. Persistence: Redis offers...
2. Data structures: Redis supports...
3. Replication: Redis has built-in...
[12 more bullets and 400 words]

Ultimately, the right choice depends on your specific requirements...
```

```
GOOD:
Redis. We already run it, and we need TTL + replication for failover.
Memcached would be simpler but we'd have to add a new service.
```

## Red Flags (Stop and Compress)

- Reply is longer than the question by 10x or more
- Reply opens with "Great question" / "Certainly" / "I'd be happy to"
- Reply has section headers in a Slack message
- Reply ends with "Let me know if you have any questions"
- Reply lists every option when the asker wanted a recommendation
- You are about to paste model output without editing it

If any of these are true: rewrite in 1-3 sentences, lead with the answer.

## Rationalizations To Reject

| Excuse                                   | Reality                                                            |
| ---------------------------------------- | ------------------------------------------------------------------ |
| "More info is more helpful"              | More text shifts the work onto the reader. Compress.               |
| "I'll cover all the edge cases"          | Asker wants your call. Offer to expand on request.                 |
| "Bullet lists are clearer"               | Bullet lists in chat are hostile. Prose fits the medium.           |
| "The AI wrote a good answer, just paste" | Pasting AI output verbatim into chat IS the slop grenade.          |
| "They might want the full reasoning"     | If they do, they will ask. Lead with the answer.                   |

## Self-Check Before Sending

1. Did I lead with the answer?
2. Could I cut this in half and lose nothing?
3. Does the length match the medium?
4. Did I remove preamble, postamble, and hedging?
5. Is this in my voice, not the model's default voice?

If any answer is no: revise before sending.