MyResearch

Claude Code skills for researchers — not developers.

For academic researchers, scientists, policy analysts, and practitioners across all disciplines.

The Problem

Most AI tools built for researchers do one of two things: search the literature or help with writing. Useful — but they stop at the surface. They don't help you decide which study design fits your question, which identification strategy to use, how to structure your theory of change, or whether your systematic review meets PRISMA standards.

The AI skill packs that do go deep — the ones that give Claude real expertise — are almost entirely built for software engineers. Git workflows, API integrations, code review. Nothing for the researcher who needs to design a mixed-methods study, write a logframe for a development donor, or run a pre-registered experiment.

MyResearch fills that gap.

What It Is

MyResearch is a collection of Claude Code skills built around the actual tasks researchers face across all disciplines — from the social sciences and humanities to the health sciences, natural sciences, and engineering. Each skill gives Claude deep, domain-specific guidance — the kind that comes from knowing the real methods, standards, and tools of a field.

It is domain-aware: on first use, you tell it which research fields you work in, and it configures itself to surface the tools most relevant to you. A public health researcher sees PICO frameworks, GRADE evidence grading, and PRISMA checklists. An economist sees DiD identification strategies, impact evaluation design, and publication-ready LaTeX tables. A development practitioner sees theories of change, logframes, and MEL frameworks.

It is open to any domain: if your field is not in the built-in list — mechanical engineering, astrophysics, urban planning, nursing — it generates a tailored skill set for you on the spot.

What It Can Do

Domain Skills

8 research domains, 4 expert skills each:

| Domain | Skills | |--------|--------| | 🏥 Public Health | Systematic review, Epi study design, Health policy brief, PRISMA flowchart | | 📊 Economics | Econometric strategy, Impact evaluation, Policy memo, Results tables | | 🌍 Climate / Environment | Impact assessment, Scenario analysis, Environmental policy brief, Adaptation framework | | 🎓 Education | Policy brief, Assessment design, Learning outcomes framework, Programme evaluation | | 🔍 Sociology / Anthropology | Field notes, Codebook, Thematic analysis, Positionality statement | | 🏛️ Political Science | Policy analysis, Stakeholder mapping, Governance framework, Comparative case design | | 🌱 Development Studies | Theory of change, Logframe, Development policy brief, MEL framework | | 🧠 Psychology | Experimental design, Scale development, Power analysis, Pre-registration |

Universal Skills

Always available regardless of domain — no configuration needed:

• Literature review and synthesis
• Grant proposal writing
• Research methodology design
• Interview and focus group guides
• Abstract writing
• Paper outlining
• Peer review response
• Research gap finding
• Journal finder

Citation Integrity

One of the most serious risks in AI-assisted research is fabricated references. AI systems — including Claude — can generate author names, journal titles, volume numbers, page ranges, and DOIs that look completely real but do not exist. A single fake reference in a submitted paper, grant proposal, or policy brief can seriously damage a researcher's credibility.

MyResearch includes a dedicated /citation-checker skill that goes beyond formatting to actively verify that your references exist.

What it does:

• Extracts every citation from your text or reference list
• Searches CrossRef, DOI registries, PubMed, and academic databases to confirm each reference is real
• Flags suspicious references before searching: impossible dates, implausible author combinations, round-number DOIs, references that seem too perfectly tailored to your argument
• Reports each reference as Verified ✅, Partially verified ⚠️ (found but details differ), Unverified ❌, or Likely fabricated 🔴
• Shows you exactly what is wrong when details don't match — it never silently corrects without telling you
• Also activates automatically when Claude itself generates text with citations, so it checks its own output before showing it to you

Supported citation styles: APA 7th edition (default), Harvard, Vancouver, Chicago, MLA, IEEE. Specify any style and it reformats your entire reference list consistently.

How to use it:

/citation-checker [paste your reference list or full paper text]

To specify a style:

/citation-checker Please use Harvard style — [paste text]

Why It Matters

Research is hard to do well. The difference between a publishable study and a flawed one often comes down to methodological decisions that are never taught explicitly — they live in the heads of senior supervisors, experienced peer reviewers, and methodologists. Most researchers, especially early-career ones, don't have reliable access to that kind of expertise on demand.

MyResearch puts that expertise inside Claude Code, available at any point in the research process — when you are designing a study at midnight, writing a grant on a deadline, responding to a reviewer's comment about your identification strategy, or checking a reference list before submission.

It is not a replacement for a supervisor or a statistician. It is the next best thing when neither is available.

Getting Started

→ Installation Guide — step-by-step setup, no technical knowledge needed

→ All Skills — complete directory of every skill with descriptions

→ Examples — practical walkthroughs for Economics, Engineering, and Public Health

→ Contributing — how to add skills, suggest improvements, or propose new domains

License

MIT — free to use, modify, and distribute. See LICENSE.