Manually invoked workflow for evidence-based PhD advisor screening, fit scoring, and tailored outreach email drafting from user-supplied applicant profiles and advisor target lists.
Use this skill when a user asks for help screening potential PhD advisors, ranking advisor fit, summarizing their applicant profile, or drafting evidence-based PhD outreach emails.
This skill is generic. It must work for any applicant across research fields. Do not assume the user's identity, field, credentials, school list, or personal documents beyond files and facts the user explicitly provides.
This skill is designed for manual invocation. The user should explicitly ask to use phd-outreach and provide:
If key inputs are missing, ask for the smallest necessary clarification or provide a fillable template.
rubrics/match_scoring.md.For every screened advisor, include:
For every drafted email, include:
Never fabricate:
When evidence is weak, stale, ambiguous, or based on only one source, mark confidence as low or medium and explain the gap.
Use phrases such as:
Avoid phrases that imply unsupported familiarity:
Personalization must be specific but restrained.
Use the 100-point rubric in rubrics/match_scoring.md.
Default grouping:
These thresholds are decision aids. Explain exceptions.
Write emails that are:
Main draft target length: 150 to 220 words.
Every email should include:
Never auto-send emails. Never instruct the model or user to mass-mail.
When screening multiple targets, use:
When drafting for one advisor, use:
Before sending any email, the user must manually verify:
If verification is not possible, keep the email more general and mark confidence low.