Future of Work Lab
← COURSES/ PERSONAL AI OS / SESSION 03
0001020304050607
50% · 4 of 8
Building Session 03
CLAUDE SKILLS FOR RESEARCH · AI for Non-Tech Students Accelerator
Build your personal AI operating system for your thesis work

Three Claude Skills your thesis will run on for the next year.

Most people use AI for research in a random way. One day they ask one kind of question. The next day they use a different process. That makes the results uneven, shallow, and hard to evaluate. Claude Skills help you turn research into a repeatable method. You can define the exact steps Claude should follow each time: how to search, compare sources, check arguments, structure findings, and produce the final output.

Train Claude on your research method once. Use it the same way every time.

Here is what most students do the day after Session 02. They open their Lit Review Project and start pasting papers in. They get decent answers, and then two problems show up. Their method drifts from one paper to the next, and they cannot get Claude to read across six papers and find the real tensions instead of surface summaries. These are not AI problems. They are research method problems.

Claude Skills fix this. A Claude Skill is a reusable procedure you give to Claude so it does the same task the same way every time. You write the procedure once in a SKILL.md file, with Claude's help, and Claude loads it on demand whenever you call it or the task matches the skill. You write it once — what the Skill does, when to use it, the exact steps Claude takes — and Claude loads it on demand across every Project. Skills are to Claude what a research methodology is to your thesis. Defined once, applied consistently.

In this session you build three Skills that run the research moves you will do most often in the early months of your thesis. You build them the easy way, using Claude's own Skill creator, which turns plain language into a working Skill in about five minutes. Once you have the format, you can build a Skill for any research move, in any field, for the rest of your career.

KEY INSIGHT
A Claude Skill is a reusable, structured set of instructions or behaviors that makes Claude perform a task the same way every time. Your method becomes a tool, not a habit you try to remember.
“A prompt asks Claude once. A Skill teaches Claude your method. Everything you read, write, and decide for the next year runs through the same procedure.”

Your method used to live in your head. Now it lives in a Skill Claude reads every time.

BEFORE · 2023 TO 2025

Every paper got a slightly different prompt.

Every time you asked AI to process a paper, you retyped the same instructions, or you skipped them and got a surface summary. Your January notes did not match your April notes because your prompt had drifted. Ten months in, your literature review was a pile of fragments you could not compare to each other.

Your method lived in your head. Your output lived in chat history. Neither one scaled.
NOW · 2026

One procedure. Every paper. Forever.

You write the procedure once in a SKILL.md file. Claude loads it on demand. Same steps, same structure, same standards, every run. The fortieth paper gets processed exactly like the first one. The Skills you build this week become a permanent part of how you work.

Your research method is now a reusable tool, not a habit you try to remember.

Three Skills. One research system.

These three Claude Skills are the early-thesis moves. The Claude Skill paper-processor turns reading into structured, comparable notes. The Claude Skill synthesis-finder is what you run every time you have four or more processed papers and need to see what they actually say together. The Claude Skill question-sharpener is what you run when your research question stops feeling sharp. Build these three first because you will use them every week. Build others later.

Skill vs. Project — why you need both. A Project is a workspace Claude reads from in every chat — your papers, your custom instructions, your knowledge base. A Claude Skill is a procedure Claude runs when you invoke it. Your Lit Review Project holds the papers; The Skill paper-processor decides how Claude reads & writes them. The Project is where your research lives. The Skill is the method Claude follows to work with it.

A quick note on slash commands. Claude has built-in tools you call by typing / in any chat — a pop-up appears with the available commands. /skill-creator is the one you'll use today to build all three Skills. You don't have to memorize the slash commands; just type / and pick from the list.

The 3 Claude Skills you'll build:

paper-processor-skill
Turn any PDF into a structured note

Every time you read a paper, you do the same thing: pull out the thesis, method, findings, limitations, useful quotes, and how it relates to your argument. Write that recipe once.

When to use: Every new paper you add to your lit review.
Input: PDF or URL.
Output: A standard note in your knowledge base with the sections you always want.
synthesis-finder-skill
Cluster what a group of papers actually says

You've read 15 papers on one sub-topic. What do they agree on? Where do they contradict? What's missing?

When to use: When you're drafting a lit-review section or mapping a debate.
Input: List of paper notes. PDFs, full articles & research papers.
Output: A synthesis with agreements, tensions, gaps.
question-sharpener-skill
Pressure-test your research question for clarity, scope, originality, and ownership

Research questions drift in the early months. This Skill names the drift — clearly — so you can decide whether to steer back or commit to the new direction.

When to use: Once a month, or before any advisor meeting about your question.
Input: Your current research question in one sentence.
Output: A PASS/FAIL/CONCERN diagnosis across four tests + one action for the week.

How to Invoke Claude Skills

  1. Next time you need that procedure, in Claude say: use the synthesis-finder skill and add your [input]. Or just say "I want this research synthesized, find the gaps" and Claude should use that Skill.
  2. Claude loads the SKILL.md, follows the steps, and returns the result.
● PRO TIP
You don't always have to name the Skill explicitly — if Claude sees a task that matches the "when to use" section, it will auto-invoke the Skill. Naming it is the safe starter move.

The Big Questions

Which research moves do you make over and over — and which one drifts most when you retype it from memory?

Consider:
  • Reading a paper, comparing two sources, pressure-testing a claim, sharpening a question. Pick the one that costs you the most when it comes out inconsistent — that is the Skill you should test first on real material in Step 5.
“The moment paper-processor and synthesis-finder exist in your account, your thesis stops being forty incomparable conversations and starts being one consistent practice.”

Claude Skills Checklist: Step-by-step Instructions

Click the bright cyan blue arrow at the end of each box to get your step-by-step instructions.

Work through each item. Check them off as you go.

0 / 6 COMPLETED
STEP 1 · Understand How Skills Work
~5 min
Learn what happens when you build a Skill — and how to invoke one when you need it
Before you build your first Skill, you need to understand two things: where Skills live, and how to call them when you need them. Five minutes of reading now saves you hours of confusion later.
STEP 2 · Build Paper-Processor
~10 min
Create a Claude Skill called paper-processor
Turns every paper you read into a structured, comparable note. This is the Skill you'll use most often — every time you add a new paper to your Lit Review. You will save 15 minutes on every research paper you read for the rest of your thesis.
STEP 3 · Build Synthesis-Finder
~10 min
Create a Claude Skill called synthesis-finder
Once you've processed 8–10 papers with paper-processor, you need a way to find patterns across them. That's what synthesis-finder does — and it's the Skill that separates a literature review from a literature summary.
STEP 4 · Build Question-Sharpener
~10 min
Create a Claude Skill called question-sharpener
Research questions shift in the early months. Students usually don't notice until they are three months deep in the wrong direction. This Skill makes question-shifts visible and intentional — not invisible and accidental.
STEP 5 · Test All Three Skills on Your Real Thesis Material
~30 min
Run all three Skills on your actual thesis work — and master both invocation styles
You've built three Skills. Now you prove they work as a system on real material, not test cases. By the end of this step, you'll have processed one real paper, run one real synthesis, and pressure-tested your real research question.
STEP 6 · Running Your Thesis Research System — Three Skills, One Pipeline
~10 min
Use each Skill in the right Project — and run them in the right order
After you have tested all three Skills, you need to know the order to use them in each Project to get the most out of them. Mixing them up is the most common way to waste a session.

What Else Can You Do with Skills

  • Extracting Key Concepts Across Papers
  • Comparing Two Papers Deeply
  • Finding Gaps in the Literature
  • Tracking Your Thinking Over Time
  • Translating Complex Papers
  • Designing Your Research Method
● UP NEXT
Next: Session 04 — Connect Your Tools

In Session 04, you'll connect your Projects to where your work actually lives — your Google Drive (for drafts and notes), your reference manager (for papers and citations), your email (for advisor correspondence), and your calendar (for deadlines). Up to now, you've been uploading files manually. After Session 04, your Projects will pull in live data automatically — and the three Skills you built today get sharper every time, because they have fresh material to work with.

← Session 02AUTO-SAVED · DRAFT