slop-patterns
Skill2 starsChanged 8 days agoMITView on GitHub
How to rewrite text a slop guard flagged, and how to keep AI tells and empty filler out of prose, commit messages, PR text, code comments, docs, chat replies, and UI. Load when a guard flags AI tells, or before drafting a commit message, PR text, a README, docs, or a landing page.
In this repository37 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- templates/CLAUDE.mdCLAUDE.md
- auditSkill
- backup-patternsSkill
- bash-patternsSkill
- depsSkill
- django-patternsSkill
- docker-patternsSkill
- drf-patternsSkill
- engineering-fundamentalsSkill
- fastapi-patternsSkill
- git-patternsSkill
- go-patternsSkill
- graphql-patternsSkill
- investigateSkill
- javascript-patternsSkill
- logging-patternsSkill
- referenceSkill
- metaSkill
- monitoring-patternsSkill
- next-app-router-patternsSkill
- nuxt-patternsSkill
- opentofu-patternsSkill
- prSkill
- python-patternsSkill
- react-patternsSkill
- report-formatSkill
- security-patternsSkill
- storybook-patternsSkill
- tailwind-patternsSkill
- test-patternsSkill
- typescript-patternsSkill
- unslopSkill
- vps-provisioningSkill
- vue-patternsSkill
- wcag-auditSkill
- writeSkill
The file
Files
Slop patterns
Two layers, kept apart:
- AI tells: surface patterns that show up far more in model output than in human writing.
- Slop: text that costs the reader more than it gives. No regexp sees it. Density, relevance, and specificity decide it.
<!-- chat: skip -->kit slop and the guards find AI tells with the lexicon in kit.yml slop:.
<!-- chat: end --><!-- chat: only
In claude.ai chat, no guard runs. The word lists and tiers are in `reference/lexicon.md` in the unslop skill.
-->Fixing the first without the second makes the text harder to spot and no better to read. A flagged sentence gets rewritten around a fact, never patched with a synonym: reference/rewrite.md.
| Writing | Read |
| Anything, first time | reference/principles.md, reference/rewrite.md |
| Prose a scan or guard flagged | reference/content-and-grammar.md |
| A commit message, PR text, a review reply | reference/commits-and-prs.md |
| Code comments and docstrings | reference/code-comments.md |
| A README or docs page | reference/docs-and-readme.md |
| A chat reply or a subagent report | reference/chat.md |
| UI: components, CSS, copy in the UI | reference/design.md |
| A hit that looks wrong | reference/false-positives.md |
| Second-guessing a rewrite as too stiff | reference/human-signals.md |
The word lists, tiers, and thresholds live only in the lexicon; these references explain them and never copy them.
Length and shape belong to rules/output.md; comment rules to rules/change.md section 5.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.