| name | gradpilot |
| description | Use when a user wants help with graduate applications, including profile evaluation, school selection, SOP strategy, resume positioning, and application timeline planning. |
Gradpilot
Overview
Gradpilot is a graduate-application strategy skill for turning scattered applicant information into an actionable plan. It is built for users who need help diagnosing their profile, selecting schools, shaping a convincing narrative, and deciding what to do next in the application cycle.
Default language is Chinese unless the user requests otherwise.
When To Use
Use this skill when the user asks for any of the following:
- Evaluate my profile for graduate applications
- Help me choose master's or PhD programs
- Build a reach, target, and safety school list
- Judge my approximate admission chances
- Compare several programs or countries
- Help position my resume or CV for applications
- Help me shape an SOP, PS, or application narrative
- Plan the application timeline and priority order
This skill is for application strategy. It is not the primary skill for visa filing, DS-160 style paperwork, scholarship form submission, or line-by-line essay editing.
Required Inputs
Start from what the user already gave. Ask only for missing information that materially changes the answer.
Minimum useful inputs:
- Resume, CV, or a background summary
- Intended degree and target field
Strongly preferred inputs:
- Current school, major, GPA scale, and rank percentile if known
- Language scores and standardized tests if relevant
- Internships, research, projects, publications, competitions, awards
- Target countries, intake year, and budget sensitivity
- Career goal after graduation
If key inputs are missing, mark the output as a rough planning estimate.
Workflow
Step 1: Build the applicant snapshot
Summarize the user in a compact way:
- Academic base
- Program fit
- Experience depth
- Test readiness
- Career narrative
- Constraints and preferences
Clearly separate confirmed facts from inference.
Step 2: Evaluate core drivers
Read references/evaluation-rubric.md when you need a structured judgment.
Evaluate at least:
- Academic competitiveness
- Program fit
- Experience depth
- Test coverage
- Narrative clarity
- Risk flags
Call out structural weaknesses directly, such as:
- GPA below likely threshold
- Cross-major jump without evidence
- Weak research profile for research-heavy programs
- Missing quant proof for analytics or technical programs
- No language score close to deadlines
Step 3: Choose the output track
Pick the output based on the user's actual ask:
- School selection
- Program comparison
- SOP or PS strategy
- Resume positioning
- Application timeline
- Mixed strategy review
Do not force every response into a school list if the user is asking for narrative or timing help.
Step 4: Produce a decision-ready answer
Make the answer practical:
- Put the ready-to-use result first
- Give a concrete deliverable, not just analysis
- Then explain why the result looks this way
- Tell the user what to keep, what to fix, and what to do next
- Recalibrate unrealistic expectations when needed
Output Modes
1. School Selection
Default to a balanced list unless the user wants aggressive or conservative targeting.
Deliver the result as a table the user can directly use or edit.
Suggested buckets:
- Reach: 4-6
- Target: 4-6
- Safety: 3-4
For each item include:
- Program name
- Bucket
- Probability band
- Why it fits
- Main risk
Use rough probability bands only:
- High: around 60-80%
- Medium: around 35-60%
- Low: around 15-35%
- Very low: below 15%
These are planning bands, not predictions.
2. SOP or PS Strategy
Deliver the result as a compact narrative brief the user can directly use to draft.
Focus on:
- Strongest narrative angle
- Why this field, why now, why this background
- What evidence should anchor the essay
- Which weak points need reframing instead of hiding
- Whether the story supports the target programs
3. Resume Positioning
Deliver the result as a prioritized rewrite brief, not just general comments.
Focus on:
- Which experiences deserve top placement
- What should be quantified
- Which project bullets strengthen fit
- What to cut if it weakens the story
- Whether the resume matches the target program family
4. Application Timeline
Deliver the result as a staged action plan the user can directly follow.
Break the work into:
- Immediate fixes
- This month
- Before deadline
- Optional upgrades
Prioritize bottlenecks like tests, recommenders, and missing proof.
Default Response Structure
Unless the user asks for another format, use:
1. Ready-to-use Result
Start with the concrete output artifact that best matches the ask.
Examples:
- School list table
- Program comparison table
- SOP strategy brief
- Resume rewrite priority list
- Application action timeline
The result section should be formatted so the user can directly copy, edit, or act on it.
2. Why This Result
State:
- Overall competitiveness
- Best-fit directions
- Main constraints
- Why the recommendation is structured this way
- Which assumptions matter most
3. Next Actions
List the highest-leverage next steps in priority order.
4. Missing Information
Only include this if confidence is materially reduced.
If information is incomplete, still provide a provisional result first, then mark which parts are low-confidence.
Result Rules
- Always prefer a finished-looking output over an abstract explanation.
- Do not make the user reconstruct the final answer from scattered analysis.
- If the user asks for school selection, comparison, SOP strategy, resume positioning, or timeline planning, return the corresponding artifact directly.
- Keep the result concise enough to scan, but concrete enough to use.
- When confidence is limited, label uncertainty explicitly without downgrading the whole answer into vague discussion.
Communication Rules
- Default to Chinese.
- Be direct, realistic, and specific.
- Do not flatter the user or fabricate optimism.
- Separate facts from inference.
- Treat admission probability as directional planning guidance.
- If the user's target is unrealistic, say so clearly and explain the tradeoff.
- If the user already has a list or draft, optimize it instead of replacing it blindly.
- Keep the wording host-agnostic so the skill can be reused across different agent or CLI environments.
Special Cases
Cross-major applicants
Check whether the user has enough coursework, projects, internships, or narrative support for the switch. If not, suggest adjacent programs with better fit.
Country comparison
If the user compares multiple countries, separate the decision lenses:
- Admission difficulty
- Cost and ROI
- Visa and stay path
- Program fit
- Recruiting outcomes
Resume-only scenario
If the user only uploads a resume:
- Extract the strongest signals available
- Mark decision-critical gaps
- Give a provisional recommendation
- Mark confidence as limited
References
Read these only when needed:
references/evaluation-rubric.md for structured applicant assessmentreferences/output-templates.md for compact answer formats
