Scholarship strategy report

Scholarship Strategy Report

You are helping a user create a professional consulting-style scholarship / study-abroad strategy report (cover page, TOC, executive summary, profile analysis, landscape, past-recipient benchmarks, comparison radar, 10 funding combinations, feasibility, roadmap, timeline, final verdict). The output is a single PDF generated by generate_report.py.

You (the agent) provide the intelligence (research, comparisons, narrative). The Python script provides the layout and charts. You collect answers, do the analysis, fill a JSON file, then run the script.

Workflow

Step 1 — Collect information (ask, do not assume)

Ask the user the questions below. Use the question tool or conversational prompts. Group them into rounds so you are not overwhelming. Never invent answers — if the user does not know, record "Not provided" and let the generator use sensible placeholders.

A. Core compare (identity & academics)

1. Full name (or preferred display name)
2. Nationality
3. Current city / state / country
4. Date of birth or age
5. Email / phone (optional, only if they want it shown)

B. Education 6. Current class / program (e.g. Class 12 State Board / CBSE / IB / Diploma / B.Tech) 7. School / college name 8. Board / university 9. Class 10 percentage / CGPA 10. Class 12 percentage (actual or predicted) 11. Expected graduation year

C. Test scores 12. SAT / ACT (score + test date) 13. IELTS / TOEFL / Duolingo (overall + subscores) 14. Olympiads, competitions, rankings, hackathons, awards

D. Projects, research & leadership 15. Research papers (titles + status: published / submitted / preprint) 16. Major projects (AI, quantum, software, hardware, etc.) 17. Companies / startups / organizations founded or led 18. Measurable impact (users, beneficiaries, countries, revenue, downloads, volunteers) 19. Positions held (captain, president, organizer, mentor)

E. Financial & logistics (drives the funding analysis) 20. Financial condition / family income bracket (this determines need-based eligibility) 21. Target country for studying abroad 22. Preferred university/college — is there one, or "no preference"? 23. Scholarship type wanted: full vs partial (if partial, what %?) 24. Scholarship sponsor preference: private company / charity / government / doesn't matter 25. Hours per week willing to work part-time, on or off campus

Step 2 — Do the analysis (your value-add)

Before writing JSON, actually think and research:

• Scholarship landscape for the target country: typical costs, what awards cover, competition levels, acceptance rates. Use web search if needed.
• Past recipient benchmarks: find 5–10 real or representative profiles of past winners for the relevant awards and extract Name / Background / Achievement / Program so the report can compare the applicant against them.
• Political & geopolitical conditions: check travel advisories, visa/permit stability, currency risk, and any active disruptions for the applicant's nationality → target country. Fold this into the Feasibility "Honest Verdict" and the conclusion.
• Comparison scoring (0–10) across dimensions: Academics, Research, Leadership, Test Scores, Community, Olympiads (or context-appropriate axes). Score the applicant ("you") and a "typical winner" for the radar chart + comparison table.
• 10 funding combinations: enumerate 10 realistic award stacks (e.g. "University Merit + Research Assistantship", "Government Bilateral + TA", "Private Sponsor + Partial"). For each, give a short description and a table with columns # | Scholarship Stack | Funding | Income | Gap | Rating, where Rating is stars (★★★★★ … ★☆☆☆☆). Exactly 10 combinations — no more, no fewer.
• Feasibility tiers (Tier 1 realistic full funding / Tier 2 strong partial+work / Tier 3 safety net) and an "Honest Verdict" with 2 numbered highlights.
• Improvement roadmap: Critical / Important / Nice-to-have, each with heading + explanation.
• 12-month timeline: months (e.g. July → June) each with checkbox tasks (□ Task).
• Final verdict bets: BEST BET, SAFEST BET, BEST RANKING, BEST SELF-FUNDING, STRETCH BET, each with a paragraph, plus a motivational conclusion.

Step 3 — Build the input JSON

Write the collected + analyzed content into a JSON file (e.g. profile.json) following the schema in INPUT_SCHEMA.md (also summarized in the appendix below). Keep narrative strings concise and objective (research-report tone: data first, then interpretation). All keys are optional; the generator fills placeholders for anything missing.

Step 4 — Generate the PDF

pip install -r requirements.txt
python generate_report.py profile.json scholarship_report.pdf

If you omit the JSON path, the script renders a built-in EXAMPLE report (useful for a dry run / demo).

Step 5 — Deliver

Give the user the PDF path and a short 2–3 sentence summary of the top recommendation and the biggest leverage action.

Design contract (must respect)

• White background, black body text, dark blue (#1F3864) headings, gray table borders, bold section titles.
• Typography: Arial/Helvetica-like. Title ~28pt, H1 ~19pt, H2 ~14pt, body 11pt, tables 9.5pt.
• Heading hierarchy: Title → Section (numbered 1–10) → Sub-heading → Paragraph → Bullets → Tables → Charts.
• Charts: radar (you vs typical winner), grouped bar (profile strength vs average), acceptance-rate bar, horizontal scholarship-value bar. The script generates these from the numeric arrays you supply — you only provide the numbers.
• Exactly 10 funding combinations in the main data section.

Appendix — minimal JSON schema

{
  "meta": { "title": "", "subtitle": "", "prepared_for": "", "date": "", "universities": ["", ""] },
  "toc": [ { "num": "1", "title": "", "desc": "" } ],
  "executive_summary": { "intro": "", "findings": ["", ""] },
  "profile": {
     "attributes": [ ["Label", "Value"] ],
     "comparison": { "headers": [""], "rows": [[""]] },
     "strength": { "labels": [""], "you": [0], "average": [0] }
  },
  "landscape": { "paragraphs": [""], "acceptance": { "labels": [""], "values": [0] } },
  "value_comparison": { "labels": [""], "values": [0] },
  "past_winners": [ { "heading": "", "description": "", "table": { "headers": [""], "rows": [[""]] } } ],
  "comparison": {
     "radar": { "labels": [""], "you": [0], "typical": [0] },
     "table": { "headers": [""], "rows": [[""]] }
  },
  "combinations": [ { "heading": "", "description": "", "table": { "headers": [""], "rows": [[""]] } } ],
  "feasibility": { "tiers": [ { "name": "", "text": "" } ], "verdict": { "highlights": ["", ""] } },
  "roadmap": { "critical": [ {"h":"", "t":""} ], "important": [ {"h":"", "t":""} ], "nice": [ {"h":"", "t":""} ] },
  "timeline": [ { "month": "", "tasks": [""] } ],
  "final_verdict": { "bets": [ {"label":"", "text":""} ], "conclusion": "" }
}

Star ratings are plain text in table cells, e.g. "★★★★☆".

See example_profile.json for a fully populated, runnable example.