Guide · Agent skill
How to use the Fabric Prompt Patterns agent skill
Updated · written with AI from the project's own documentation and checked against our catalogue data
The Fabric skill gives an AI agent a library of about 240 ready-made prompt recipes from the Fabric project. It covers summarizing, extracting insights, analyzing logs or claims, writing security findings and making diagrams. After this guide you will know what the skill does, how to install it, how to ask for a pattern in plain language, and what to watch for on safety.
Fabric Prompt Patternsdanielmiessler/LifeOS · MIT Download the skill Quick start
- 1Download the ZIP. Download the Fabric skill ZIP from this page.
- 2Unzip the skill. Unzip the file so the Fabric folder is in place.
- 3Place the folder. Move it to ~/.claude/skills/ for personal use, or to .claude/skills/ in a project.
- 4Or upload it in the apps. In the Claude apps, go to Settings → Capabilities → Skills and upload the skill.
- 5Make a natural request. Ask for a pattern by name or describe the task, and paste your input text.
- 6Check the output. Confirm the result matches the pattern's format and verify key facts against your source.
What the Fabric skill makes the agent do
An agent skill is a folder of instructions that an AI agent can load when a request matches it. The Fabric skill adds a set of named prompt patterns. A pattern is a tested prompt format, such as a recipe for turning an article into a structured summary or a set of threat scenarios from a system description.
When you ask for a pattern by name or by what you want done, the agent reads the matching recipe and applies it directly to your text. It does not call an outside program for this. The output keeps the format the pattern defines.
The skill also includes refined versions of the most popular patterns, and the agent uses those in place of the originals. Some requests, such as word summaries like ws 3, are handled by a dedicated intent in the skill.
Install the Fabric skill
Download the Fabric ZIP from this page and unzip it. For a personal setup, place the unzipped skill folder in ~/.claude/skills/. For a single project, place it in that project's .claude/skills/ folder.
You can also upload the skill in the Claude apps under Settings → Capabilities → Skills.
- Personal install: ~/.claude/skills/
- Project install: .claude/skills/ inside your project
- Claude apps: Settings → Capabilities → Skills, then upload
Trigger the skill with natural requests
You do not need special syntax. Name the pattern or describe the job, and the agent will match the request to the skill. Requests that the skill lists as triggers include phrases such as "summarize", "analyze logs", "threat model", "commit message", "explain code" and "sigma rules".
Be specific about the input. Paste the text, attach the file or give the URL, and say what you want back.
- "Use fabric extract_wisdom on this article: [paste text]"
- "Analyze these logs and tell me why the deploy failed."
- "Write a commit message for this diff."
- "Create a threat model for this login service using STRIDE."
- "ws 3 on this paragraph" for a word summary at three levels
Examples of what you can ask for
The skill covers several groups of work. Writers and researchers can ask for summaries, extracted ideas, or an analysis of a paper's claims. Security engineers can ask for threat scenarios, a pentest finding write-up, a Sigma or Semgrep rule, or an email header analysis. Developers can ask for a pull request description, a code review of a design document, or a Mermaid diagram.
Expect the output to follow the format of the named pattern, so the same kind of request gives a consistent structure each time.
What the skill needs and how it runs
Most patterns run with no extra tools. The skill runs scripts and uses the network. The Bun runtime is needed only to run the CheckPatterns.ts checker in the evals section. The Fabric CLI is needed only for YouTube transcripts (-y) and for blocked URLs (-u).
The skill includes evaluation files for its ported patterns, such as AnalyzeLogs and AnalyzeClaims, plus a checker script that verifies key pattern files exist.
- Bun runtime: only for the CheckPatterns.ts checker
- Fabric CLI: only for YouTube transcripts and blocked URLs
- Network access: used by the skill
Safety notes before you paste sensitive material
The text you give the skill is processed by the model. Do not paste material you would not send to an AI service.
Summaries of URLs or of text of 200 characters or more also trigger a separate harvest skill that writes a KNOWLEDGE note. Review that behaviour before relying on the skill for URL summaries.
The output is only as accurate as the model's reading of the source. Check important facts against the original.
- Do not send confidential or regulated data unless your policy allows it
- Review the harvest behaviour for URLs and long text
- Verify facts, figures and quotes against the source
Tips for better results
Give the full source text rather than a short description, and name the pattern you want. If the agent picks the wrong recipe, restate the pattern name explicitly.
For multi-agent investigations or deep research, use a different skill. The Fabric skill excludes those, and also excludes content-adaptive extraction, which belongs to ExtractWisdom.
- Name the pattern directly when you know it
- Paste complete source text, not summaries of it
- Use another skill for deep research or ExtractWisdom-style adaptive extraction
What people use it for
- Turn a long article into a structured summary with extracted insights
- Draft threat scenarios or a STRIDE threat model for a system design
- Analyze a set of server logs to find the likely cause of a failure
- Write a commit message or pull request description from a code change
- Create a Mermaid diagram or a Sigma rule from a described behaviour
Frequently asked questions
How do I install the Fabric skill?
Download the ZIP from this page and unzip it. Place the folder in ~/.claude/skills/ for personal use or in a project's .claude/skills/ folder. You can also upload it in the Claude apps under Settings → Capabilities → Skills.
What do I need to run the Fabric skill?
Most patterns need no extra tools. The Bun runtime is needed only to run the CheckPatterns.ts checker, and the Fabric CLI is needed only for YouTube transcripts (-y) and blocked URLs (-u).
Is the Fabric skill free?
The skill is published under the MIT licence, as listed on this page. The source does not state pricing for other services, so check any model or service costs separately.
Is it safe to send sensitive documents to the Fabric skill?
Be cautious. The text is processed by the AI model, so do not paste material you would not send to an AI service. Summaries of URLs or long text also trigger a harvest skill that writes notes, so review that behaviour first.
What is the difference between Fabric and ExtractWisdom?
The Fabric skill runs prompt patterns, and it excludes content-adaptive extraction, which is handled by ExtractWisdom. Use Fabric for named patterns and ExtractWisdom for adaptive wisdom extraction.
Why did the agent not use the pattern I expected?
The skill maps some pattern names to refined versions of the originals, and it runs those instead. If the wrong recipe is chosen, restate the exact pattern name in your request.
New agent skills, repos and AI tools — every week
The best new skills and open-source projects, the fastest-growing repos, and one free prompt. Five minutes to read. Unsubscribe anytime.
Our verdict
Use the Fabric skill if you want consistent, structured summaries, analyses and security write-ups from proven prompt formats, and you are comfortable with the text going to an AI model. Pick something else for multi-agent investigations or deep research, or when the material is too sensitive to share with an AI service.
Download the skill