InfraNodus Claude Skills: Extended Prompts for Your AI Workflows

Claude Skills is a relatively new feature available in Anthropic LLM clients that allows you to write the extended prompts with code blocks to perform specific tasks or follow a certain logic. The models automatically select which skills to use depending on the context of the conversation.

For example, if you have a "cognitive variability" skill from InfraNodus installed in your Claude Web client, as soon as you tell the model that you're stuck with a certain idea, it will automatically invoke that skill and use the underlying methodology described in the skill to help you develop your idea further. It will also know which InfraNodus MCP server tools to invoke to help you with your task.

Skills are like compact descriptions of custom logics that augment AI capabilities with your own methodologies and workflows. If you're a philosopher or a researcher, it's a great way to prime your model to think in a certain way. If you're a marketer, you can use skills to adhere to your brand's values and style.

InfraNodus has a collection of skills that can be used to augment your AI workflows with our methodologies based on cognitive variability, text network analysis, auto-generation of knowledge graphs and ontologies, and constructing and analyzing social and text networks. You can use the skills in conjunction with the InfraNodus MCP Server to perform more complex tasks and workflows.

You can also use the InfraNodus MCP Server to develop your own skills and make sure that your prompts cover all the most important use cases and don't have any gaps.

Access our Skills Collection - includes installation instructions

Use Cases for the InfraNodus Claude Skills

Here are some examples of the Claude Skills from our collection that can be used to augment your AI workflows:

Cognitive Variability - A skill that promotes the cognitive variability framework for adaptive thinking to help evolve ideas and avoid getting stuck.
Critical Perspective - Engage in critical thinking by questioning assumptions, exploring alternative perspectives, and uncovering latent topics in conversations.
Writing Assistant - Grammar correction, style refinement, and pattern detection that can trigger cognitive state analysis.
Ontology Creator - Build structured ontologies, taxonomies, and knowledge graphs by extracting entities, defining relationships, and mapping conceptual structures for any topic or for a collection of texts.

Note, that usually skills can also contain code blocks that can be executed but we prefer to comnbine them with MCP servers instead to avoid exposing API keys and for better workflow and version control.

Example: Cognitive Variability Schema

InfraNodus promotes cognitive variability across two spectrums: scale and intent. It nudges a thinking process to alternate between exploration and focus, small scale and big scale, and to connect and disrupt ideas. This is done through the graph interface, workflow scenarios, and instructions that encourage the user to switch attention, intention, and scale.

We built this schema both into our main graphical interface, into the MCP server, and into the Claude Skills collection. The schema works best with the InfraNodus MCP server as it has access to network analysis tools that help diagnose the state of a discourse and recommend the next actions.

Use Case: Developing the Content of an Article

To demonstrate how the InfraNodus skills work, we will demonstrate how we can use it on this very article. Considering that the "cognitive-variability" skill is already installed in Claude, we copy all the previous text (up to this moment) and ask Claude to help us develop the ideas further:


												I feel stuck with this text, can you help me develop it further:
												... [pasted text] ...

Normally, without any skills installed, Claude would respond with generic suggestions, maybe ask a few questions, try to extrapolate. That would be hit or miss.

Claude immediately recognizes the trigger pattern—"stuck" signals a specific cognitive state. It silently reads the cognitive variability skill, understands you're likely in a "Biased" state (tunnel vision on one angle), and applies the framework without you asking. The most amazing thing is that it could also be your own methodology, which is very useful for building custom workflows:

Claude then provides us with the advice that includes the gaps we could focus on and some practical steps to develop the text in this article further:

In our case, it suggested us to add more practical examples ("show, don't tell") and thus diverify the content of the article and make it appeal to a broader audience.

We can then ask Claude to help us refine the text and make it more engaging and appealing to the audience:


											I like the first two points about making it more practical (points 1 and 2). Can you propose something I could write in my style on that topic further?

Claude automatically identifies the writing assistant skill and helps us refine the text in our style:

In our case, we added this very section following the suggestion of Claude where we provide a description of a practical example. I don't like to copy / paste AI-written text (even if it's written in my style), so I rewrote it, but Claude definitely helped to understand what should be the next topic to focus on.

We can then use the InfraNodus MCP server to help us refine this text further through bridging the content gaps and making it more SEO friendly:

The SEO report from the InfraNodus MCP tool tells us to focus more on the practical aspects and some technical problems that people might have: installing the skills and also addressing the cases when they are not invoking. Which we're going to do in the next section.

Installing Claude Skills

Claude skills will work with any LLM client, but Claude has a more automatic support for them:

Installing Skills to Claude Web / Desktop

To install the Claude Skills to Claude Web / Desktop client, you need to:
1) go to the Settings > Capabilities,
2) activate the "Code execution and file creation"
3) scroll down to the "Skills" section and activate some of the default skills you find useful
4) you can also download the InfraNodus skills collection and upload each individual skill (.zip or .skill file) by clicking "Upload Skill"

Installing Skills to Claude Code

In this case, you will need to manually move each skill's folder (e.g. `skill-cognitive-variability`) to the `~/.claude/skills` directory in your home directory. In this case, the skills will be available for all your projects.

If you want the skill to be available in a specific project, you need to create the `.claude/skills` folder in that project and copy the skills folders there.

Installing Skills to ChatGPT and other LLM clients

To install the skills to any other LLM client, you can upload it to the Project's instructions / specific chat where you want to use them or create a Custom GPT in ChatGPT to use that skill (using the content of the SKILL.md file). Then you can invoke the still in that particular project / chat, or if it's a custom GPT, you can invoke it with the @skill_name mention.

Claude Skills FAQ

Here are some frequently asked questions about Claude Skills:

What to do when the skills are not invoking?

If your particular skill is not invoking automatically, you can
1) ask Claude to invoke it explicitly: "Use the .... skill to ...."
2) add a custom instruction to your general user settings or project settings to invoke that particular skill
3) improve the description of the skill and reupload it to Claude
4) temporarily disable the MCP tools that tend to have priority over skills if that causes the issue

What's the difference between using the skills vs the MCP server?

Skills are a more lightweight way to invoke a specific logic or workflow. They are more suitable for high-level tasks and for invoking specific tools and workflows. MCP servers are more comprehensive, offer better control of the tasks, and are better suited for connecting Claude to external tools, databases, and services.

What's the difference between the Claude Skills and ChatGPT Custom GPTs?

Claude Skills are very similar to Custom GPTs, but their advantage is that they are more customizeable, can have a modular structure, and used in different LLM clients.

The main difference, however, is how they are invoked. With Custom GPTs in ChatGPT, you have to explicitly call the Custom GPT, while with Claude Skills, they can be automatically invoked by Claude depending on the context of the conversation. Claude uses so-called "progressive loading" where it first ingests the skill's description and name and only then gets into the detailed description of the skill once it decides to use it.

How many skills should I use?

There's a hard limit of about 20 skills that can be installed in Claude and it's better to stick with this number. Choose your skills carefully and the less you have installed, the better they will be selected by Claude.

Improve Your Skills with the InfraNodus MCP Server

You can use the InfraNodus MCP Server to develop and improve your skills. In this case, once the skill is created in Claude, activate the InfraNodus MCP server and ask it to find gaps in the skill's content.

Alternatively, you can also use the InfraNodus BrowserExtension to develop and improve your skills. In this case, open the skill's document and then open the extension and choose "Analyze Document" in its settings to detect the gaps and generate interesting questions (prompts) that will help Claude address some of the blind spots in the skill's description.

Our skills collection is also aware of the various InfraNodus tools available, so installing the InfraNodus MCP server will also improve the quality of the skills' output. You can try the MCP server for free, but to avoid rate limits and get extended functionality

About InfraNodus