Match a deep student research profile to professors, chairs, researchers, and recent papers using the professor seed index plus native web research. Use when asked to find proposal-relevant advisors, rank chairs or supervisors, compare matches, explain fit, or prepare evidence for precise research proposals without a database or UI.
설치
Codex 또는 Claude로 설치 이 Prompt를 복사해 Codex, Claude 또는 다른 어시스턴트에 붙여 넣으면 Skill 페이지를 검토하고 설치를 진행할 수 있습니다.
Match a deep student research profile to professors, chairs, researchers, and recent papers using the professor seed index plus native web research. Use when asked to find proposal-relevant advisors, rank chairs or supervisors, compare matches, explain fit, or prepare evidence for precise research proposals without a database or UI.
Match Thesis Advisors
Rank possible thesis advisors by combining the student's in-session research profile with the professor seed index and evidence gathered from official web sources. Treat the ranking as input to research-proposal generation, not as the final product.
Workflow
Build or reuse a deep student profile. If the user only gave broad interests or one short answer batch, use build-student-profile first and ask the next 3-5 coaching questions.
Read ../find-university-chairs/references/professors/INDEX.md for professor names and official starting URIs.
Use find-university-chairs and find-recent-papers to gather current public evidence with the active agent's native websearch/browser tools.
Score candidates qualitatively by proposal fit, research taste fit, evidence freshness, prerequisite fit, and risk.
Return a ranked shortlist with reasons, caveats, backup options, and proposal hooks.
Output
For each match include:
chair/lab and relevant person
fit summary
evidence from official pages or publication sources
matching papers or research areas
prerequisites and preparation
risks or caveats
proposal hooks that could become research questions
suggested next action
End with 2-3 backup directions if the top matches are too competitive or uncertain.
Rules
Do not produce advisor rankings from a shallow profile or after only one brief question batch. Ask profile-building questions first.
Do not invent supervision capacity, open topics, quotas, team sizes, citations, or willingness to supervise.
Distinguish "strong research fit" from "confirmed available thesis topic".
Do not treat old bundled chair, researcher, or paper profiles as the primary source.
If websearch/browser tools are unavailable, say that advisor matching can only use the professor seed names and URIs until the user provides page contents or enables browsing.
Do not depend on the old UI, backend API, database, Docker, Celery, or FastAPI app.
