Agent skill
de-slop
Remove LLM-isms and AI writing patterns from text. This skill should be used when editing prose to sound less like AI output — removing overused words, fixing structural tells, and restoring natural human voice. Triggers: "de-slop", "remove AI writing", "humanize this", "sounds too AI", "LLM-isms", "AI slop", or when reviewing text that reads like chatbot output.
Install this agent skill to your Project
npx add-skill https://github.com/petekp/agent-skills/tree/main/skills/de-slop
Metadata
Additional technical details for this skill
- author
- petekp
- version
- 0.1.0
SKILL.md
De-Slop
Strip AI writing patterns from text to restore natural, human-sounding prose.
Based on Wikipedia: Signs of AI writing and WikiProject AI Cleanup.
When to Use
- Editing any prose that sounds like chatbot output
- Reviewing drafts generated with AI assistance
- Self-check before publishing AI-assisted writing
- When text feels "off" but the reason is hard to pinpoint
Process
Step 1: Diagnose
Read the full text before changing anything. Load references/word-list.md and references/structural-patterns.md to identify which patterns are present.
Categorize findings into three severity levels:
Red — Immediate tells (fix first)
- Chatbot leakage ("I hope this helps", "Certainly!", template blanks)
- Grandiose filler ("stands as a testament", "in today's fast-paced world")
- Synonym cycling (same entity referred to by 4+ different names)
Yellow — Statistical signals (fix in clusters)
- 3+ words from the overused word list appearing in close proximity
- Rule of three used more than twice
- Tailing participle phrases ("emphasizing the significance of")
- Em-dash density higher than ~1 per 200 words
Green — Structural patterns (require rewriting, not word swaps)
- Relentless balance (every section same length)
- Uniform register (no tonal variation)
- Generic specificity (hypothetical examples, no real names)
- Excessive hedging (qualifiers every third sentence)
- Risk aversion (no specific claims, no edge)
Present the diagnosis as a brief summary before making changes. Example:
Diagnosis: 4 red flags (chatbot leakage, grandiose filler), 7 yellow signals
(word clusters in paragraphs 2, 5, 8), 2 green patterns (relentless balance,
uniform register).
Step 2: Fix Red Flags
Remove or replace all Red items. These are unambiguous AI artifacts.
Chatbot leakage: Delete entirely.
Grandiose filler: Replace with plain statements or delete.
- "stands as a testament to" -> "shows" or "is"
- "plays a vital role in shaping" -> "shapes" or "affects"
- "in today's fast-paced world" -> delete (it never adds meaning)
Synonym cycling: Pick one term and stick with it. Use pronouns for variety.
Step 3: Fix Yellow Signals
Work through clusters. The goal is not to ban specific words but to break up detectable patterns.
Word clusters: Replace overused words with plain alternatives.
- "delve into" -> "look at" / "examine" / (often just delete)
- "leverage" -> "use"
- "robust" -> "strong" / "solid" / (ask: is this adjective needed at all?)
- "nuanced" -> "detailed" / "complicated" / (often delete)
- "landscape" -> name the actual domain
- "multifaceted" -> drop it; describe the actual facets instead
- "crucial" / "pivotal" / "paramount" -> "important" or delete
Copula avoidance: Restore simple verbs.
- "serves as" -> "is"
- "features" / "offers" / "boasts" -> "has"
Transition abuse: Remove mechanical connectives.
- "Moreover," / "Furthermore," / "In addition," -> start the sentence without them, or use "and" / "also"
Rule of three: Break at least half of them. Use two items, or four, or one.
Tailing participles: Rewrite as separate sentences or delete.
- "..., emphasizing the importance of X" -> delete, or: "X matters because..."
Step 4: Fix Green Patterns
These require actual rewriting, not substitution.
Relentless balance: Redistribute weight. Expand important sections. Trim or collapse unimportant ones. A 3-sentence paragraph next to a 12-sentence paragraph is fine.
Uniform register: Inject tonal shifts. A blunt short sentence after a complex one. A casual aside in a technical passage. Let the writing breathe.
Generic specificity: Replace hypothetical examples with real ones, or remove examples that add nothing.
Excessive hedging: Remove qualifiers that don't reflect genuine uncertainty. If something is true, state it without "often" / "generally" / "can be."
Risk aversion: Sharpen claims. Add an opinion. Allow an imperfect sentence to stand if it has energy.
Enthusiasm gap: Vary paragraph investment. Spend more words where the writer (or subject) is more interesting.
Step 5: Final Read
Read the entire edited text once more. Check for:
- Overcorrection — Did fixes make the text choppy or too informal? Restore where needed.
- Meaning preservation — Does every sentence still say what it originally meant?
- New patterns — Did edits introduce their own repetitive patterns?
- Voice consistency — Does the text sound like one person wrote it?
Principles
- Prefer plain words. "Use" over "leverage." "Is" over "serves as." "Important" over "crucial."
- Prefer short sentences. Break long compounds. Not every thought needs a clause.
- Preserve meaning. Never change what the text says, only how it says it.
- Don't over-correct. Some em dashes are fine. An occasional "furthermore" is fine. The goal is to break patterns, not ban words.
- Real > hypothetical. A named example beats "consider a scenario where..."
- Uneven > balanced. Spend more words on what matters more.
- Specific > vague. "Response time dropped from 200ms to 50ms" beats "significantly improved performance."
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
multi-model-meta-analysis
Synthesize outputs from multiple AI models into a comprehensive, verified assessment. Use when: (1) User pastes feedback/analysis from multiple LLMs (Claude, GPT, Gemini, etc.) about code or a project, (2) User wants to consolidate model outputs into a single reliable document, (3) User needs conflicting model claims resolved against actual source code. This skill verifies model claims against the codebase, resolves contradictions with evidence, and produces a more reliable assessment than any single model.
capture-learning
Analyze recent conversation context and capture learnings to project knowledge files (for project-specific insights) or skills/commands/subagents (for cross-project patterns). Use when the user asks to "capture this learning", "update the docs with this", "remember this for next time", "document this issue", "add this to CLAUDE.md", "save this knowledge", or "update project knowledge". Also triggers after resolving build/setup issues, discovering non-obvious patterns, or completing debugging sessions with valuable insights.
optimize-agent-docs
Build a retrieval-optimized knowledge layer over agent documentation in dotfiles (.claude, .codex, .cursor, .aider). Use when asked to "optimize docs", "improve agent knowledge", "make docs more efficient", or when documentation has accumulated and retrieval feels inefficient. Generates a manifest mapping task-contexts to knowledge chunks, optimizes information density, and creates compiled artifacts for efficient agent consumption.
agent-changelog
Compile an agent-optimized changelog by cross-referencing git history with plans and documentation. Use when asked to "update changelog", "compile history", "document project evolution", or proactively after major milestones, architectural changes, or when stale/deprecated information is detected that could confuse coding agents.
literate-guide
Create a narrative guide to a codebase or feature in the style of Knuth's Literate Programming — code and prose interwoven as a single essay, ordered for human understanding rather than compiler needs. Use when the user asks to 'explain this codebase as a story', 'write a literate guide', 'create a narrative walkthrough', 'tell the story of this code', 'Knuth-style documentation', 'weave a guide for this feature', or when they want deep, readable documentation that treats the program as literature. Also trigger when someone wants a document that a thoughtful reader could follow from start to finish and come away understanding both WHAT the code does and WHY every design choice was made.
autonomous-agent-readiness
Assess a codebase's readiness for autonomous agent development and provide tailored recommendations. Use when asked to evaluate how well a project supports unattended agent execution, assess development practices for agent autonomy, audit infrastructure for agent reliability, or improve a codebase for autonomous agent workflows. Triggers on requests like "assess this project for agent readiness", "how autonomous-ready is this codebase", "evaluate agent infrastructure", or "improve development practices for agents".
Didn't find tool you were looking for?