How to Add Skills to Claude AI: Text Humanizer  🔗 **Source:** [How to Add Skills to Claude AI: Text Humanizer](https://www.youtube.com/watch?v=GsLk-9QrS54#transcript) > [!note]+ Document Notes > tags: [claude-ai, ai-skills, text-humanization, github, developer-tools, no-code] > author: [MISSING] > source: Visualmodo > reference: https://www.youtube.com/watch?v=GsLk-9QrS54 > date created: 2026-06-18 11:00 > > # 🏷️ Title: How To Add Skills to Claude AI: Text Humanizer > <!-- deep-read:sections start --> ## 🧭 Table of Contents - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📚 References](#-references) - [🗣️ Paraphrase](#️-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-assessment) - [✅ Logical Validity](#-logical-validity) - [🛡️ DARVO Analysis](#-darvo-analysis) - [🎯 Conclusion](#-conclusion) ### 🔗 References - [Claude AI](https://claude.ai) - [GitHub — search "Claude AI Humanizer"](https://github.com/search?q=claude+ai+humanizer) - [Anthropic Agent Skills documentation](https://docs.anthropic.com/en/docs/agents-and-tools/agent-skills/overview) ## 📝 Summary - **What happened.** A ~2-minute Visualmodo tutorial walks through installing an Agent Skill into Claude AI's web app, using a publicly available "Humanizer" skill found on GitHub as the worked example. The host downloads the skill's `.zip`, opens Claude's Personalize/Customize → Skills panel, uploads the file via drag-and-drop, and toggles it on. - **Why it matters.** Skills let non-developers extend Claude with packaged instructions (a folder of Markdown plus optional resources) without writing integration code. The video is a concrete, screen-by-screen demo of the upload-a-skill path. - **Key stakeholders.** Claude AI end users, content creators wanting to "humanize" AI-generated text, and anyone building or distributing Claude Code / Claude AI skills as `.md` or `.zip` packages. - **Risks and tradeoffs.** "Humanizing" AI text raises authenticity and disclosure concerns; the demo installs a third-party skill from GitHub without any vetting guidance. No mention of reviewing skill instructions for safety before enabling. - **Bottom line.** Adding a skill to Claude AI is a download → upload → toggle flow inside the Skills panel; the same panel also lets you create, browse, edit, replace, or uninstall skills. ## 📖 Main Content - **What a "skill" is here.** In Claude's product, a Skill is a packaged set of instructions (and optional files) that Claude loads to perform a specialized task on demand. The video treats it as a drop-in extension — no API or coding required. - **The example skill.** The host searches GitHub for "Cloud AI Humanizer" (transcribed from "Claude AI Humanizer") and picks a public repository described as a humanizer for AI-generated text. The repo includes a README with usage instructions. - **Obtaining the package.** From the repo, use the green **Code** button → download the `.zip`. The skill ships as files Claude can ingest directly. - **Where skills live in Claude.** In the Claude AI left sidebar, open **Personalize / Customize**, then **Connectors and Skills**, then **Skills**. The panel includes a **skill creator** (with a guide), a **search** for browsing skills, and a **plus (+)** button to add a skill. - **Three ways to add a skill.** (1) Create one yourself by writing instructions, (2) let Claude help create it, or (3) **upload** an existing skill as `.md` or `.zip`. The host chooses upload as "the easier way." - **Installing it.** Click drag-and-drop, select the downloaded skill, open it; the skill appears as installed and "working." A toggle in the upper-right turns the skill on or off, and you can view its instructions by code or as normal text. - **Managing it later.** The three-dots menu (upper-right) exposes replace, edit, try, and **uninstall**. The video frames removal as equally simple. - **Uncertainties / contested claims.** The transcript is auto-captioned and garbles "Claude" as "Cloud" and the skill toggle as "Q." The specific repository name ("Blader Humanizer") is likely a transcription artifact and should be verified before trusting any named repo. No safety/vetting step is shown. ### 🗒️ Notes #### ⏱️ [00:00:00](https://www.youtube.com/watch?v=GsLk-9QrS54&t=0) Intro and the GitHub search - Host (Louder, Visualmodo WordPress themes) frames the goal: install skills on Claude AI and humanize text with one. - Standard subscribe/like/share ask. - Direction: search GitHub for "Claude AI Humanizer" and find the public humanizer repo with instructions. #### ⏱️ [00:00:33](https://www.youtube.com/watch?v=GsLk-9QrS54&t=33) Download the skill from GitHub - Read the repo's README instructions first. - Use the green **Code** button → scroll down → download the `.zip`. - Then open Claude AI and find **Personalize / Customize** in the left sidebar → **Connectors and Skills** → **Skills**. #### ⏱️ [00:01:09](https://www.youtube.com/watch?v=GsLk-9QrS54&t=69) The Skills panel and add options - There's a **skill creator** with a guide for building your own. - Next to the search icon, a **plus (+)** button adds a skill. - You can browse existing skills, create with Claude, write instructions, or **upload** a skill as `.md` or `.zip`. Host picks upload. #### ⏱️ [00:01:42](https://www.youtube.com/watch?v=GsLk-9QrS54&t=102) Upload and enable - Click drag-and-drop → select the downloaded skill → open. The Humanizer skill is now installed. - A toggle in the **upper-right corner** turns the skill on/off. - View instructions by code or as normal text. #### ⏱️ [00:02:13](https://www.youtube.com/watch?v=GsLk-9QrS54&t=133) Uninstall and outro - To remove: upper-right **three dots** → options to replace, edit, try, or **uninstall**. - Closing subscribe/comment ask; "see you guys tomorrow." ## 📚 References - [Claude AI — web app](https://claude.ai) - [Anthropic — Agent Skills overview](https://docs.anthropic.com/en/docs/agents-and-tools/agent-skills/overview) - [GitHub search: Claude AI humanizer skills](https://github.com/search?q=claude+humanizer+skill) - [Visualmodo](https://www.visualmodo.com) ## 🗣️ Paraphrase - The host introduces a quick tutorial on installing skills in Claude AI and using one to humanize AI-written text, then asks viewers to subscribe. - He directs viewers to search GitHub for a public "Humanizer" skill, read its instructions, and download the project as a `.zip` via the green Code button. - In Claude AI, he opens the Personalize/Customize menu, navigates to Connectors and Skills, and selects Skills. - He notes the panel offers a skill creator and a plus button to add skills, with options to browse, create with Claude, write instructions, or upload a `.md`/`.zip`. - He uploads the downloaded skill by drag-and-drop, and it appears installed and working, with a toggle to enable or disable it and a way to view its instructions. - He closes by explaining that the three-dots menu lets you replace, edit, try, or uninstall the skill, then signs off. ## 📰 Newsletter **Teach Claude a new trick in two minutes** If you've been hearing about Claude "Skills" and wondering how the upload actually works, this short Visualmodo walkthrough is a tidy demo. The recipe: grab a public skill from GitHub (here, an AI-text "humanizer"), download the `.zip` from the green Code button, then in Claude AI head to Personalize → Connectors and Skills → Skills, hit the plus, and drag-and-drop your `.md` or `.zip`. Flip the toggle and it's live; the three-dots menu handles edit, replace, and uninstall. For WordPress folks who tinker with content workflows, it's a gentle on-ramp to packaging your own repeatable instructions as a skill — the same mechanic that powers Claude Code skills. One caveat the video skips: a skill is just instructions someone else wrote, so read the README before you enable a stranger's package. And "humanizing" AI text is a tool with ethics attached — use it to polish, not to deceive. ## ✅ Action Items #### DevRel - Developer advocacy: Not applicable to this source — consumer how-to with no advocacy program. - Education: produce a short "anatomy of a Claude skill (`.md`/`.zip` structure)" companion explainer for non-developers. - Developer experience: document the upload-vs-create-vs-Claude-built skill paths with screenshots. - Developer marketing: Not applicable to this source. - Community management: Not applicable to this source. - Internal/external/cross-project: note that the Claude AI upload flow mirrors Claude Code skill packaging — cross-link guidance. #### OSPO - Policy & Governance: Not applicable to this source. - License Compliance: flag that GitHub-sourced skills carry licenses; check before redistributing. - Security Management: Not applicable to this source — though "vet third-party skill instructions before enabling" is a relevant guardrail. - Software Inventory: Not applicable to this source. - Developer Support: Not applicable to this source. - Community Engagement: Not applicable to this source. - Strategic Alignment: Not applicable to this source. #### Community - Maintainers: Not applicable to this source. - Contributors: Not applicable to this source. - Sponsors: Not applicable to this source. - Ecosystem: Not applicable to this source. #### WordPress - Core: Not applicable to this source. - Core Performance: Not applicable to this source. - Core AI: Not applicable to this source. - **Core Program Team:** Not applicable to this source. - Mobile: Not applicable to this source. - Playground: Not applicable to this source. - Community: Not applicable to this source. - Training: a "build your first Claude skill" mini-lesson would suit WP content creators. - Marketing: Not applicable to this source. - Support: Not applicable to this source. - Design: Not applicable to this source. - Documentation: Not applicable to this source. - Photos / TV / Openverse: Not applicable to this source. - Polyglots: Not applicable to this source. - Accessibility: Not applicable to this source (see Accessibility Assessment for the video's own gaps). - Meta: Not applicable to this source. - CLI: Not applicable to this source. - Hosting: Not applicable to this source. - Tide / Themes / Plugins: Not applicable to this source. #### The WPCC - Programs: Not applicable to this source. - Projects: Not applicable to this source. - Funding experiments: Not applicable to this source. ### FAIR - Technical independence: Not applicable to this source. - Policies and covenants: Not applicable to this source. - Productized governance artifacts: Not applicable to this source. ## 🧪 Hypotheses - H1: A drag-and-drop `.zip` upload flow lowers the barrier to skill adoption versus requiring users to author skill instructions from scratch. - H2: Skills sourced from public GitHub repos without a vetting step increase the risk of users enabling unsafe or low-quality instructions. - H3: "Text humanizer" skills measurably alter AI-detector classification of generated text. ### 📊 Methodology - H1: A/B usability test comparing time-to-first-working-skill for upload vs. author-from-scratch cohorts. - H2: Audit a sample of public "humanizer" skills for prompt-injection or data-exfiltration patterns in their instructions. - H3: Run identical prompts with and without the humanizer skill through common AI-text detectors; compare scores. ### ❓ Questions - Does Claude AI sandbox or review uploaded skill instructions before they run? - What is the exact, current navigation path (Personalize vs. Customize) given UI changes over time? - Which public repository is the host actually using, given the garbled "Blader Humanizer" transcription? ### 🔮 Future Research - Comparative review of skill-distribution mechanics across Claude AI, Claude Code, and competing assistants. - Ethics of AI-text "humanization" tools and detection arms race. - Best practices for vetting third-party skill packages before enabling. ### 💡 Implications - **Practice:** Packaging repeatable instructions as a skill is now a no-code workflow for end users. - **Policy:** Organizations should set guidance on which third-party skills employees may enable. - **Security:** Uploaded skills are executable instructions; treat them like any third-party dependency. - **Education:** Short procedural demos like this are effective for the "where do I click" gap, but should pair with a vetting note. - **Funding models:** Not applicable to this source. ### 🧠 Learning Styles Assessment - Reading level estimate: spoken auto-caption transcript; ~Flesch–Kincaid Grade 6–8 (short, imperative sentences, common vocabulary). Flesch Reading Ease ~70–80. Caption noise inflates apparent errors. #### 👀 Visual learners - Present: strong — it's a screen-share walkthrough with on-screen clicks (Code button, Skills panel, drag-and-drop, toggle, three-dots). - Improve: add labeled callouts/arrows and a one-frame flow diagram (download → upload → toggle). #### 🔊 Aural learners - Present: narration throughout. - Improve: cleaner audio (the host clears his throat repeatedly); a 30-second spoken recap of the steps. #### 🗣️ Verbal learners - Present: plain step-by-step narration; key terms named (skill, `.md`, `.zip`, toggle, uninstall). - Improve: define "skill" and "humanizer" explicitly up front; on-screen glossary. #### 🧩 Kinesthetic learners - Present: strong — it's a do-along procedure with concrete clicks. - Improve: a downloadable checklist and a sample skill to practice with. #### 🧠 Logical learners - Present: clear linear sequence (find → download → navigate → upload → enable → manage). - Improve: a decision tree for the three add paths (create / Claude-built / upload). #### 🧑‍🤝‍🧑 Social learners - Present: comment-section invitation for questions. - Improve: a linked community thread or Discord for skill-sharing. #### 🧘‍♀️ Solitary learners - Present: self-paced, repeatable steps. - Improve: a written transcript/blog version to follow alone. #### 🌿 Naturalistic learners - Present: none. - Improve: tie to a broader "ecosystem of reusable skills" framing. ### 📊 Accessibility Assessment #### Screen reader semantics - Present: YouTube provides a video element and (auto) captions. - Improve: ensure the description/page repeats the steps as text; on-screen UI labels are not screen-reader exposed in video. #### Keyboard navigation - Present: Not applicable to video content itself; the YouTube player is keyboard-operable. - Improve: verify player controls expose visible focus and labels. #### Contrast and readability - Present: depends on screen-share UI; Claude's panel generally meets contrast norms. - Improve: zoom into clicked controls so low-vision viewers can follow. #### Text resizing and reflow - Present: Not applicable to video; viewers can fullscreen. - Improve: provide a text version (blog) that reflows at 200% zoom. #### Media and captions - Present: auto-captions exist but are inaccurate ("Cloud" for Claude, "Q" for the toggle, "Blader" for the repo) — fails WCAG 1.2.2 quality expectations. - Improve: human-corrected captions and a downloadable transcript; avoid relying on auto-captions for a procedural tutorial. #### Forms and interactions - Present: Not applicable to video content. - Improve: Not applicable. #### Standards reference - Maps to WCAG 2.2 AA: 1.2.2 Captions (auto-captions inaccurate — needs correction), 1.2.5 Audio Description (UI clicks not narrated for non-sighted users), 2.4.6 Headings/labels (provide a text outline of steps). ### 🌍 DEIB Assessment #### Cultural breadth - Present: single host, single example; English only. - Improve: subtitles in additional languages; note that GitHub/Claude UI may differ by locale. #### Inclusive language - Present: friendly, plain, no exclusionary language ("What's up, guys" is casual but informal). - Improve: a more gender-neutral greeting than "guys"; define jargon (`.md`, `.zip`) for newcomers. #### Neurodiversity support - Present: short, chunked steps suit limited working memory. - Improve: on-screen step numbers and a written checklist for those who prefer reading. #### Representation - Present: one voice/perspective. - Improve: Not strongly applicable for a 2-minute solo tutorial. #### Psychological safety - Present: low-pressure, encouraging tone; invites questions in comments. - Improve: add a brief "it's okay to undo — here's uninstall" reassurance early (the video does cover uninstall at the end). ### ✅ Logical Validity #### Strong reasoning - Procedural/sequential: the steps form a coherent, verifiable cause→effect chain (download the package → upload it → it appears installed → toggle enables it). This is demonstrative rather than argumentative, and the demonstration is internally consistent. - Analogical: implicitly treats a skill as a drop-in file, which matches Claude's actual `.md`/`.zip` skill packaging. #### Fallacies - **Appeal to ease / oversimplification** — "all you have to do is..." and "the easier way" present uploading a third-party skill as trivially safe, omitting any vetting step. Location: [00:00:33] and [00:01:09]. Fix: add a "review the skill's instructions before enabling" caution. - **Unsupported claim** — "There is amazing skills right here" / "It's a great public hit" asserts quality without evidence. Fix: state selection criteria. #### Tone of voice - Casual, upbeat, encouraging, promotional (subscribe/like/share bookends). Instructional and reassuring; not argumentative. ### 🛡️ DARVO Analysis - Not applicable to this source. This is a neutral procedural tutorial with no accusation, conflict, or defensive posture. There is no Deny, Attack, or Reverse-victim/offender content to analyze. ## 🎯 Conclusion - **Bottom line.** Adding a skill to Claude AI is a short, no-code loop: find a skill (here a GitHub "humanizer"), download its `.zip`, open Claude's Personalize → Connectors and Skills → Skills panel, click plus, drag-and-drop the `.md`/`.zip`, and toggle it on. The same panel handles create, edit, replace, and uninstall. - **Next steps for readers.** (1) Read any third-party skill's README/instructions before enabling it. (2) Try the create-with-Claude path to author your own skill. (3) For repeatable workflows, package instructions as a skill rather than re-typing prompts. (4) Treat "humanizing" AI text as an editing aid, not a deception tool. ## 🔄 Refactor Appendix — How To Add Skills to Claude AI: Text Humanizer > Density: 2–5 Literature atoms (H3), 0–3 Permanent candidates (H4). > Short procedural source — minimal atoms. > > Tag pool: #bmf · #addendum · #xref · #articleseed · #storystem · #share · #aphorisms ### 📚 Literature Atoms (H3) #### [H1] Claude AI installs skills by uploading an .md or .zip > [!quote] ID:rw_pending > "You can upload in .md or .zip, okay? Click on drag and drop files, select the skill that you just have downloaded, click on open." **Context & use**: - [00:01:09]–[00:01:42]: Personalize/Customize → Connectors and Skills → Skills → plus → upload. - Mirrors Claude Code skill packaging; useful reference for how-to-distribute-a-skill mechanics. **Tags**: #xref **Highlight**: [view highlight](readwise://pending) <!-- atomicity: 1 idea ✓ --> #### [H2] A skill toggle and three-dots menu manage the skill lifecycle > [!quote] ID:rw_pending > "click into these three little dots and uninstall this Q. You can replace, you can edit, you can try and a shot or uninstall." **Context & use**: - [00:01:42]–[00:02:13]: enable/disable toggle (upper-right) plus replace/edit/try/uninstall. - "Q" is an auto-caption mangling of the skill toggle; verify naming against live UI. **Tags**: #xref **Highlight**: [view highlight](readwise://pending) <!-- atomicity: 1 idea ✓ --> ### 🧠 Permanent Candidates (H4) None worth promoting — the source is a single procedural demo with no reusable cross-project claim that survives the atomicity + reusability filter. ### ✏️ Fleeting Queue - [F1] Verify the actual GitHub repo behind the "humanizer" (transcript garbles it as "Blader Humanizer") before recommending it — https://www.youtube.com/watch?v=GsLk-9QrS54 #fleeting #inbox - [F2] Open question: does Claude AI vet uploaded skill instructions before they run? #fleeting #inbox ### → Things 3 (max 2 actions) - [ ] Note the Claude AI upload-skill flow (`.md`/`.zip` drag-and-drop) in the Claude Code skill-building reference — #things <!-- deep-read:sections end --> 📄 **Full text:** ![[Full Document Contents/Articles/How to Add Skills to Claude AI Text Humanizer.md|How to Add Skills to Claude AI: Text Humanizer by Visualmodo]] ## Highlights - We have a skill creator, okay? There is a guide for creating skills. You can create your own, but next to the search icon, there is this plus icon to add skill. Click on it. You can browse the skills in the case that you want. There is amazing skills right here. Or you can create a skill. You can create with Cloud. Cloud can help you to create the skill. You can write the skill instructions or upload a skill. I'm going to upload the skill, the easier way. You can upload in .md or .zip, okay? ([View Highlight](https://read.readwise.io/read/01kvdcrhzvmyeg9dgf8baq2j3z)) - Note: H1 — core mechanic: add a Claude skill via the + icon; browse, create-with-Claude, write instructions, or upload an .md/.zip (relevant to building Claude Code skills). - Tags: atom/literature/h1, claude-skills - section below in the case that you have any question. Don't forget that in the case that you would do not like this Q and want to delete, you can have you the upper right corner of your screen, click into these three little dots and uninstall this Q. You can replace, you can edit, you can try and a shot or uninstall. And that is. Hope you guys enjoy this video. Feel free to use the comment section below in the case that you have any question. And I'll see you guys tomorrow. All the best. ([View Highlight](https://read.readwise.io/read/01kvdcrnmnhr5er8vketpdkfy0)) - Note: H2 — skill lifecycle management: toggle, then replace / edit / try / uninstall via the three-dots menu. - Tags: atom/literature/h2