mohamedhadyashry/medical-research-thesis-supervisor

This project is a Claude Agent Skill, a set of instructions that changes how Claude behaves within a conversation, in this case turning it into something closer to a strict academic research supervisor rather than a general writing assistant. It is aimed at people writing medical theses, dissertations, systematic reviews, or journal papers who need real scientific rigor rather than fluent but unreliable text. The core promise is that it will not make up citations or data. If Claude cannot actually find and verify a real source using a search tool, the skill has it say so plainly and ask the user for sources rather than inventing a plausible looking reference, a fabrication problem the README says is common with general-purpose AI writing help. The skill walks a project through several stages: helping refine a vague research idea into a real, checkable question, building structured literature searches across academic databases, weighing the quality of different types of studies rather than treating them equally, writing in a plain scientific style without marketing language or overstated claims, and then reviewing its own draft the way a critical peer reviewer would before submission. It can also track progress across multiple sessions, so a long-running project does not need to be re-explained each time. To use it, you install the skill files into Claude, either by uploading a zip file through the Claude.ai website, copying the folder into Claude Code's skills directory, or loading it through the Claude API for developers. It only works well if web search or a research database connection is turned on, since without a way to look things up, the skill deliberately refuses to write citations from memory. The README is explicit that this tool is meant to support genuine, disclosed academic work, not to help disguise AI-written text as entirely human-written, and that users remain responsible for verifying and standing behind every claim in their own work.