AI/Skills/competition-strategy-kit
Past papers into a plan.
Feed it any contest's or standardized exam's past papers, and it returns a priority-tiered "how to score more" strategy guide — the key facts per theme, the most common trap, a year-by-year question map — and, where a categorised book build already exists, a corrected, compiling prep book to match.
competition-strategy-kit is subject-agnostic — math, physics, chemistry, biology, CS/USACO, economics, verbal — and format-agnostic — multiple-choice, short-answer, written proof, or mixed. It reads a body of past papers, categorises the questions by theme, finds where the marks actually are, and writes a tiered guide: what to master first, the trap examiners set most often, and a map of how each theme recurs year to year. When the project already carries a categorised-book build, it also produces the corrected, compiling prep book — the same pipeline behind Tian2's printed solution volumes.
In
- A set of past papers for one exam (any subject, any format)
- Optionally an existing categorised-book build to correct against
- Target language(s) for the guide
Out
- A priority-tiered "how to score more" strategy guide
- Key facts per theme + the most common trap
- A year-by-year question map
- Optionally a corrected, compiling categorised prep book
A theme map, tiered by where the marks are
Given several years of a physics contest's papers, the skill returns a tiered theme map. The fragment below is illustrative of the shape of that output.
/competition-strategy-kit ./past-papers/physics-challenge/
--years 2016-2024 --lang en
Kinematics & energy conservation. Appears every year, usually as the opening two questions. Common trap: forgetting to check the sign convention after a bounce. Master these to bank the reliable marks.
Circuits & thermal physics. Recurs in 7 of 9 years. Common trap: treating internal resistance as negligible when the question hints otherwise.
Modern & astrophysics. Sparse and hard — the marks that separate the top decile. Worth it only once Tiers 1–2 are secure.
Themes, weights, and traps shown are illustrative of the output structure.
This is the analysis front-end to Tian2's book pipeline: the strategy guide tells a student where to spend their hours, and the categorised-book side connects to the same LaTeX tooling that produces the printed solution books and their online question banks.