Claude Skills
Overview
Skill Use Cases
Reference

Writing Assistant Skill for Claude & AI Agents


The Writing Assistant skill (skill-writing-assistant) refines texts in any language: it fixes grammar, spelling, and punctuation, paraphrases only when a sentence is unclear, and keeps the result sounding like you wrote it — not an AI. It preserves your markdown, links, formatting, and intentional stylistic quirks, and answers in the same language as your input.

It's made for writers, bloggers, researchers, and anyone editing texts with an LLM who is tired of getting back homogenized, obviously AI-flavored prose. Beyond editing, it has a second layer: it reads your grammatical patterns as signals of how you're thinking — and tells you when the structure of your text needs attention, not just the surface.

Get Started

1
Create an Account

Sign up for a free InfraNodus account to get access to the knowledge graph tools and your API key.

Create a Free Account
2
Get InfraNodus MCP
https://mcp.infranodus.com

Add this URL as a connector via Settings > Connectors in Claude Web / Desktop and authenticate with your InfraNodus API key via OAuth.

3
Install the Skill
/plugin marketplace add infranodus/skills
/plugin install skill-writing-assistant@infranodus-skills

Run these two commands in the Claude Code prompt to install the skill as a plugin (use infranodus-all@infranodus-skills to install all our skills at once). Or use npx in your terminal: 

npx skills add infranodus/skills --skill skill-writing-assistant -g -a claude-code -y

npx comes bundled with Node.js. For Claude Web / Desktop, ChatGPT, and OpenClaw, see the full installation instructions.

Didn't work? You can simply download skill-writing-assistant.zip from our GitHub repository and add it to your LLM client manually — see the step-by-step instructions.

What the Skill Does

Once installed, prompts like these trigger the skill:

  • "Fix the grammar in this text" — corrects errors and returns only the refined text, no changelog, no "here's the corrected version" preamble.
  • "Make this sound less like AI" — removes AI-detection red flags: "Moreover", "It's important to note that", formulaic structures, too-neat parallel constructions.
  • "Edit this, but keep my voice" — minimal intervention: it changes what needs changing and leaves working sentences alone, including creative punctuation like tildes and em-dashes.
  • "Check if this article is balanced" — analyzes the topical structure and flags when one topic dominates unintentionally (while respecting deliberate emphasis in manifestos or poetry).
  • "Help me develop this text" — for longer texts, identifies underdeveloped topics, missing connections, and content gaps against real search demand.

The output rule is strict: just the refined text, ready to paste. Explanations appear only when you ask for strategic development, where the content gap findings are the point.

How It Works

The core is a set of editing principles: preserve the author's voice, intervene minimally, keep all formatting intact, and match the input language and formality. The skill carries an explicit list of AI red flags to strip — from transitional-phrase overuse to excessive hedging — and an equally explicit list of what not to do: don't add transitions that weren't there, don't homogenize sentence length, don't trade personality for "professional" blandness.

The distinctive part is the pattern-detection layer. While correcting, the skill acts as a sensory system: grammatical patterns are treated as signals of cognitive states and structural problems.

  • Repetitive sentence structures — possible fixation on one idea; the text needs diversification.
  • Error clustering in one section — unclear thinking there; that section needs development.
  • Missing transitions — structural gaps between ideas that need bridging.
  • Tense inconsistency — temporal imbalance: stuck in the past, present, or future.
  • Pronoun ambiguity — concepts that need explicit naming and defining.
  • Punctuation rhythm — short choppy sentences suggest drilling down (possibly obsessively); long flowing ones suggest exploring; question-mark clusters suggest dispersed thinking.

These patterns then act as triggers. When they suggest a structural issue, the skill escalates from grammar fixing to analysis with the InfraNodus MCP tools (if connected): a text overview to check topical balance, develop_text_tool to find underdeveloped topics and content gaps in problem sections, and generate_seo_report to compare your text against real Google search demand. The feedback you get combines the corrected text with pattern-based insights, e.g. "repetitive sentence structures in paragraph 3 suggest this concept needs varied exploration".

The MCP tools are optional and reserved for substantial texts (500+ words) that need content strategy — for everyday grammar fixes the skill works standalone, with no MCP server required.

Install the Skill

Claude Code

Download the skill into your global skills folder:

cd ~/.claude/skills
curl -L -o skill-writing-assistant.zip https://github.com/infranodus/skills/releases/latest/download/skill-writing-assistant.zip
unzip skill-writing-assistant.zip -d skill-writing-assistant

Then just paste a text and ask Claude to refine it — the skill activates on editing requests.

Claude Web / Desktop

Go to Settings > Capabilities, enable Code execution and file creation, scroll to the Skills section, and upload the skill-writing-assistant.zip file.

OpenClaw

Run in the OpenClaw chat:

install this skill: https://github.com/infranodus/skills/releases/latest/download/skill-writing-assistant.zip

In OpenClaw, the skill can also auto-configure the InfraNodus MCP connection via mcporter using your INFRANODUS_API_KEY. See the main installation guide for step-by-step instructions per LLM client.

What's Next

Get the InfraNodus MCP