Skip to content

Stop Slop — Setup Guide

Source: hardikpandya/stop-slop GitHub: hardikpandya/stop-slop Skills: 1 skill · 10.6K installs Category: Content Quality First Seen: catalogued August 17, 2026 sweep (stop-slop on skills.sh since January 20, 2026) Quality Tier: 🟢 Production — Gen Agent Trust Hub Pass, Socket Pass, Snyk Pass (3/3)

Stop Slop eliminates predictable AI writing patterns from prose: filler phrases, formulaic structures, passive voice, vague declaratives, and narrator-from-a-distance framing. Its seven core rules read like a compressed style guide for human-sounding output — and it even bans em dashes, matching our own public-content rule.

Installation

npx skills add hardikpandya/stop-slop

Individual skills install with the explicit repo form:

npx skills add https://github.com/hardikpandya/stop-slop --skill stop-slop

Prerequisites

Requirement Details
Node.js + npx For the skill installer
Prose to clean Any draft — post, email, doc, or page copy

What It Provides

The skill's seven core rules:

  1. Cut filler phrases, emphasis crutches, and adverbs
  2. Break formulaic structures (binary contrasts, negative listings, rhetorical setups)
  3. Active voice — every sentence has a human subject doing something
  4. Be specific — no vague declaratives or lazy extremes
  5. Put the reader in the room — "you" beats "people"
  6. Vary rhythm — mix sentence lengths; no em dashes
  7. Trust readers — state facts directly, skip hand-holding

Reference files (references/phrases.md, references/structures.md) carry the concrete pattern lists.

Quick Start

  1. Install: npx skills add hardikpandya/stop-slop
  2. Ask: "run stop-slop on this draft"
  3. Apply the seven rules; the pattern references guide the rewrite

CorpusIQ Use Cases

Use Case How
Public content gate Pairs with our content-voice rules and slop scoring — a second pass before anything ships publicly
Email quality Executive-facing and outreach emails read human at first pass
UGC script cleanup Video scripts and captions lose the AI rhythm before recording

Limitations / Verification

  • Security audits on stop-slop: Gen Agent Trust Hub Pass, Socket Pass, Snyk Pass — clean across all three
  • Publisher-page install counts verified (10.6K single skill); GitHub 15.7K stars on the repo
  • Single-skill cluster — the whole value is the seven rules plus the two reference files
  • Complementary, not competitive, with avoid-ai-writing: stop-slop is the compressed ruleset; avoid-ai-writing is the audited multi-mode workflow
npx skills add hardikpandya/stop-slop   # verify install works

Powered by CorpusIQ