hero problem produces how compare install results sample
AI-agent skill · MIT Lic· consulting-grade

A strategy report
that thinks like an advisor.

An AI-agent skill that turns a structured applicant profile into a 16-20 page, chart-rich scholarship & study-abroad strategy PDF Report — with 10 ranked funding combinations, quantified cost gaps, past-recipient benchmarks and an honest verdict on whether your goal is even realistic.

10
ranked funding combos
117+
countries tested
100%
validity score
16-20
page Report PDF
agent · scholarship-strategy-report
python
skills.sh
agentskill.sh
agent chat
The problem

Search is scattered.
A skill is decisive.

A normal web search returns a wall of links and scattered facts. It never stacks awards into a funding path, never quantifies the gap you still owe, and never tells you the one thing that matters: is this goal even possible?

Without the skill
Dozens of links, zero ranked pathways — you assemble the plan yourself.
No quantified cost gap; the family's real exposure stays invisible.
No benchmark against past recipients — you can't tell if you're competitive.
Invalid-premise goals survive: an impossible target looks achievable.
Advice quality scales with your own research skill and English — widening the gap for the disadvantaged.
With the skill
Exactly 10 ranked, star-rated funding combinations — each with a quantified net gap.
Residual cost the family must cover is computed, not guessed.
You vs the typical winner, scored on a radar across 6 dimensions.
An explicit honest verdict: impossible / conditional / selective / consistent.
Reproducible & machine-verifiable — same profile yields the same report.
What it produces

A consulting-grade report,
13 sections, chart-rich.

Every run emits a single polished PDF: cover, table of contents, executive summary, profile analysis, landscape, past-recipient benchmarks, comparison radar, the 10 funding combinations, feasibility, roadmap, a 12-month timeline and a final verdict.

01
Executive Summary
Intro plus the key findings and the single highest-leverage action.
02
Applicant Profile
Attribute table, 5-university comparison and a strength bar chart.
03
Scholarship Landscape
Costs, coverage and competition with an acceptance-rate bar chart.
04
Past Recipients
Benchmark case-study tables of real prior awardees.
05
Comparative Analysis
You vs typical winner — radar chart plus a scored table.
06
Funding Combinations
Exactly 10 ranked, star-rated award stacks with gaps.
07
Feasibility
Tier rankings plus an honest, two-point verdict.
08
Improvement Roadmap
Critical / important / nice-to-have, each with a rationale.
09
12-Month Timeline
Month-by-month checklist from intake to program start.
10
Final Verdict
Best / safest / ranking / self-funding / stretch bets + conclusion.
11
Radar & Bar Charts
Rendered with matplotlib — you supply numbers, it draws.
12
Political Risk
Travel advisories, visa stability & currency risk folded into the verdict.
How it works

Five loops. One report.

The skill is a versioned, reproducible workflow built on a five-stage loop-engineering framework. Each loop is independently testable — and an ablation shows loops L2–L4 produce ~69% of the decision-grade artifacts.

L1
Retrieval
Pull the scholarship landscape, costs and past-recipient data for the target country.
L2
Contradiction Detection
Flag impossible or conditional goals first — e.g. English-medium medicine in Austria.
L3
Quantification
Stack awards, compute funding, income and the residual net gap per combination.
L4
Synthesis
Rank 10 combinations, benchmark vs winners, build the radar and timeline.
L5
Verification
Programmatically confirm the PDF is valid and carries the L2 verdict.
Why different

Built for decisions,
not for browsing.

Capability
This skill
Web search
Consultancy
Ranked funding pathways
Yes — 10 combos
No
Partial
Quantified cost gap
Yes — per combo
No
Sometimes
Past-recipient benchmark
Yes — radar
No
Manual
Honest goal verdict
Yes — L2
No
Rare
Reproducible output
Yes — versioned
No
No
Cost
Free · MIT
Free
High $
Install & use

Three ways to run it.

Works with 70+ agents (opencode, Claude Code, Cursor, Codex…) via skills.sh, or run the generator directly with Python.

skills.sh
agentskill.sh
Python
# One command — works with 70+ agents
$ npx skills add piyushomanwar16/scholarship-strategy-report-skill

# Then just ask your agent for a scholarship strategy report
"Build me a funded-study strategy for an MSc in Canada"
# Via agentskill.sh
$ npx @agentskill.sh/cli@latest install \
    @piyushomanwar16/scholarship-strategy-report-skill

# Or import the repo at https://clawhub.ai/import
# (OpenClaw / ClawHub)
# Local / manual — clone, then run the generator
$ pip install -r skills/scholarship-strategy-report/requirements.txt
$ python skills/scholarship-strategy-report/generate_report.py \
      skills/scholarship-strategy-report/example_profile.json demo.pdf

# Omit the JSON to render the built-in EXAMPLE report
Empirical results

Measured against normal search.

From the accompanying research paper: a controlled A/B test of 10 matched experiments (9 real profiles across Austria, Germany, UK, USA, Canada, Australia, France, Netherlands, Switzerland + one sparse profile). Both arms got the identical prompt — so every gap is attributable to the synthesis loop, not to better inputs.

100%
validity score
vs 0% for web search
100%
decision-grade coverage
vs ≈10% baseline
10
matched experiments
across 9 countries
69%
of artifacts from
loops L2–L4

Decision-grade coverage

% of 7 decision-critical artifact classes produced

Goal-validity verdict

correctly classifying goal as impossible / conditional / selective / consistent

Inside a sample report

The charts, rendered live.

These four charts are drawn from the bundled example_profile.json (a Canada data-science case). The Python generator produces the same figures in the PDF via matplotlib.

You vs the typical winner

comparative radar · 6 dimensions

Profile strength

you vs applicant average

Acceptance rate by pathway

% conversion for well-matched profiles

Scholarship value index

relative funding value (higher = better)

View a generated sample report

The full 16- 20 page Report PDF built from the example profile — every section, table and chart described above. Generated by generate_report.py.

example-report.pdf →

Read the research paper

The controlled empirical comparison with conventional web search: methodology, per-experiment case studies, ablation and the L5 verification pass.

research_paper.pdf →