# Deslop Audit > Deslop Audit is an AI agent skill for Claude Code, Codex CLI, Gemini CLI, and other agents that read SKILL.md. Scores a draft for machine tells, then reviews it, interviews you, or rewrites it clean. Best for: humanizer skill for Claude. - URL: https://atharvashah.com/ai-skills/deslop-audit - Price: $49, one payment - Works with: Claude Code, Codex CLI, Gemini CLI, Cursor, GitHub Copilot - Category: Writing and voice - Built by: Atharva Shah (https://atharvashah.com) - Requirements: Needs a paid Claude plan with Claude Code and Python 3 for the scorer. It works on English prose. It does not promise to pass any third-party AI detector. - Buy: https://buy.polar.sh/polar_cl_R6zXD453Fqls2A2MvMzI0peVQkdv2aw1TZP2c1jBqZo?custom_field_data.skill-premium=deslop-audit (the skill folder and setup guide arrive by email) ## Take the machine tells out of your draft, and keep you in. It scores your draft for the patterns readers now spot as AI, then reviews it, rewrites it, or interviews you for the facts it lacks. It never invents a number to make a sentence sound better. ## What it produces - **Fix mode, line by line.** Each machine pattern gets rewritten in place, with the claim kept and the tell named. ![A real LinkedIn draft in fix mode: three struck-through AI patterns, each with its plain rewrite and a pattern tag.](https://atharvashah.com/ai-skills/deslop-audit/scene-1.webp) - **The reader audit.** Review mode tells you which patterns readers catch, and how often, before you touch a word. ![A real review report on 78 published posts, listing five systemic AI patterns with how many posts each one hit.](https://atharvashah.com/ai-skills/deslop-audit/scene-2.webp) - **A post ready to ship.** You get back a finished post with the same story, steps, and call to action. ![A real LinkedIn post after fix mode, with the rewritten sentences highlighted in yellow.](https://atharvashah.com/ai-skills/deslop-audit/scene-3.webp) ## Pick a mode, and it does only that job. Review changes nothing. Interview asks before it writes. Fix rewrites and shows you every change. 1. **Point it at the draft.** One file, a list of files, or a whole folder. The scorer runs first because it is fast, free, and does not guess. 2. **Read the diagnosis.** It leads with the largest problem, then gives a verdict per passage: keep, revise, ask you, or cut. A clean draft gets a short report that says so. 3. **Approve the rewrite.** Fix mode keeps every claim, keeps the length close, and reports each pattern it changed. It scores the new text before it hands it over. ## Most AI checkers only hand you a percentage. A detector tells you a number. A humanizer app rewrites your facts along with your tells. ### Deslop Audit - **Named findings.** Every hit names the pattern and the line it sits on. - **Three modes.** Review, interview, or fix, and review never edits. - **No invented facts.** A gap becomes a bracket or a question for you. - **Runs on your machine.** Drafts stay in your folder next to your Claude session. ### AI detector sites - **A number, no fix.** You learn the draft reads as 71 percent AI and nothing else. - **No line to act on.** They rarely point at the sentence that caused the score. - **Scores drift.** Paste the same text twice and the number can move. - **Your draft on their server.** Unpublished work goes to a site you do not control. ### Humanizer apps - **Synonym swaps.** They trade one tell for a stranger word and call it human. - **Facts drift.** A careful claim comes back stronger than your evidence. - **Your voice sanded off.** Odd details and uneven rhythm are the first things cut. - **Word credits.** A long article eats a month of credits in one paste. ## Frequently asked questions ### What is a humanizer skill for Claude? A humanizer skill is a SKILL.md folder that teaches Claude to find and remove the patterns readers spot as AI writing. This one scores the draft first, names every pattern and the line it sits on, then rewrites only what it flagged. ### How do I remove AI writing patterns from a draft? Run the skill in review mode to see each pattern with its line number, then in fix mode to rewrite them. It cuts filler, breaks repeated sentence shapes, and turns vague claims into brackets you fill with your own facts. ### Will it make my writing pass AI detectors? It removes the patterns readers and detectors react to, but no tool can promise a detector score. The goal is prose a person trusts. ### Which mode should I start with? Review. It changes nothing and tells you whether the draft needs a fix or an interview. When you name no mode, it runs review. ### Does fix mode change my facts? No. Every claim survives, and a missing number turns into a visible bracket for you to fill. It checks the rewrite for new facts before you see it. ### Can I add my own banned words? Yes. The list is a plain file. Add a word or a phrase and the scorer blocks it from the next run. ### Does it work on a whole folder? Yes. It ranks the files by the findings a reader would notice and fixes them one at a time, so each file keeps its own voice. ### Which Claude plan do I need? A paid plan with Claude Code. Pro covers normal use. A large folder sweep can hit usage limits and wait for a reset. ### What else do I pay for? Nothing. The scorer is a local Python script and needs no API key. ### Does it work for docs and emails, or only essays? It judges each piece by its job. A README does not need a bold opinion, and an email keeps its request. ### What if the first rewrite feels flat? Paste two or three things you wrote before AI tools, and it matches that rhythm. It runs one self-review and stops, because more passes flatten prose. ### Does this skill work with Gemini CLI, Codex, Cursor, or GitHub Copilot? Yes, with one caveat. The skill is a SKILL.md folder in the Agent Skills open standard, which Claude Code, Codex CLI, Gemini CLI, Cursor, and GitHub Copilot all load. It is built and tested on Claude Code. Steps that lean on Claude-only features, such as parallel subagents, run one after another or need a manual step in other agents. ### How do I install an agent skill? Copy the skill folder into your agent's skills directory: ~/.claude/skills for Claude Code, or the skills folder your agent documents for Codex CLI, Gemini CLI, Cursor, or Copilot. The agent reads the description and loads the skill when your request matches it. Each skill ships with a setup guide for any tool it needs. [All AI skills](https://atharvashah.com/skill-pack) ยท [Library index](https://atharvashah.com/ai-skills.md)