Humanize KR (im-not-ai)
epoko77-ai/im-not-aiA Korean-language humanizer that detects and rewrites AI-sounding text (translationese, mechanical parallelism) while preserving facts, with deterministic gates.
Recommended
Check first Depth 4/5
Stop Slop Claude skill
A single writing skill that tells the agent to strip common AI tells from prose and score the result on five dimensions before delivering it.
Our verdict Recommended with caveats
A clear, compact anti-slop checklist with useful references, but some rules are absolute (no adverbs at all, no passive voice, no Wh- openers) and will over-correct some kinds of writing unless you loosen them.
What Stop Slop does
Stop Slop is one SKILL.md of about 2.6 KB that the agent applies when drafting, editing or reviewing prose. It sets eight core rules. The agent cuts filler phrases and all adverbs, breaks formulaic structures such as binary contrasts, negative listings, dramatic fragments and rhetorical setups, uses active voice with a human subject, names specifics instead of vague declaratives, addresses the reader as "you," varies sentence rhythm, skips hand-holding, and rewrites anything that reads like a pull-quote. Em dashes and sentences that open with a Wh- word are also banned.
Before delivering, the agent runs a quick checklist: adverbs, passive voice, inanimate things doing human verbs, "here's what" openers, "not X, it's Y" contrasts, three sentences of matching length, and punchy closing one-liners. It then rates the draft 1 to 10 on directness, rhythm, trust, authenticity and density, and revises anything below 35 of 50. Three reference files load on demand: references/phrases.md lists phrases to remove, references/structures.md covers structural patterns, and references/examples.md shows before and after rewrites. The README also describes using the files in Claude Projects, in custom instructions or in an API system prompt.
Good for
- Editing blog posts, essays and newsletters that read as machine-written
- A final pass on drafts before publishing
- Writers who want an explicit checklist and score instead of a vague request to sound human
Not for
- Technical or reference docs where passive voice and adverbs are sometimes the clearest choice
- Non-English text
What's inside depth 2/5
One SKILL.md of about 2.6 KB with three short reference files of phrases, structures and examples, and no scripts, tests or evals.
- Skills
- 1
- Reference docs
- 5
- Scripts
- 0
- Tests / evals
- 0
- Templates
- 0
- Avg SKILL.md
- 2.6 KB
Counted from the repository file tree. How we score depth.
Security review
No issues foundWe scanned all 7 files in this repository and read every script. Nothing in it reaches beyond what the skill describes.
This skill is a set of writing-style instructions for removing predictable AI patterns from prose. It contains no code, makes no network calls, and does not touch files outside the text you ask it to edit.
What we checked
The repository (7 files) on Oct 29, 2026. We had every file and read every script.
NVIDIA SkillSpector static scan: 0 pattern hits. SkillSpector reported no hits. The repo holds only Markdown files. The domains it mentions (hvpandya.com, github.com) appear as author and attribution links, not as endpoints the skill calls.
A review, not a certification. Skills change after we check them — read the files before installing.
License
MIT, commercial use allowed.
Can it run unattended?
Not really. It is an editing pass applied to each draft, with no loop, state or stop condition beyond the 35/50 revise threshold.
Alternatives to Stop Slop
Anti-Defensive Writing
Adkid-Zephyr/anti-defensive-writing-SkillA lightweight CN/EN academic-writing skill that restructures paper drafts around one strongest claim instead of hedged, evenly-weighted self-critique.
With caveats
No issues Depth 2/5
LinkedIn Marketing Skills
sergebulaev/linkedin-skills12 skills that draft LinkedIn posts, comments, replies and profile copy in your voice, read engagement data via Apify, and publish through Publora after you approve.
With caveats
Check first Depth 3/5
FAQ
What is Stop Slop?
Stop Slop is a skill by Hardik Pandya that teaches Claude or another LLM to find and remove predictable AI writing patterns. It combines banned phrases, structural patterns to avoid, sentence-level rules and a five-part scoring rubric.
How do I install Stop Slop?
In Claude Code, add the repo folder as a skill. In Claude Projects, upload SKILL.md and the reference files to project knowledge. For API calls, the README says to include SKILL.md in the system prompt, with the reference files loaded on demand.
Is Stop Slop safe to install?
The repo contains only Markdown files and a license. The automated scan found no pattern hits and no executable files.
Does Stop Slop work for technical writing?
You can use it, but some rules are strict: it bans all adverbs, all passive voice, em dashes and sentences that start with a Wh- word. Expect to relax a few of these rules for documentation, where passive voice or a precise adverb can be the clearer choice.
Sources: GitHub repository (stars, license, file tree, last push Mar 17, 2026). Review based on the README and SKILL.md files as published.