What problem does it solve?
This Skill helps ICPC problem production reliably create multiple wrong solutions and the corresponding generators/data so that near-TL or subtly incorrect approaches are caught during local verification.
Core Features & Use Cases
- Mistake-driven wrong-solution enumeration: Systematically covers “idea mistakes,” “complexity mistakes,” and “implementation mistakes” using the provided mistake taxonomy.
- Multi-family wrong solution implementation: Produces several compilable
src/wrong/*.cpp programs plus a src/generator/generator.cpp that can generate targeted cases.
- Config-aligned data generation: Updates
config.json fields tied to generator.cases and wrongSolutions so generated datasets effectively distinguish the intended wrong families.
- Targeted and stressful test design: Mixes directed datasets (to defeat specific wrong families) with random data as supplements, ensuring enough scale and strength to hit performance and edge traps.
- Interactive-aware wrong strategies (if applicable): Extends enumeration to query strategy, order, and answering/fallback behaviors for interactive problems.
Quick Start
Ask the agent to produce wrong solution families and a generator for your current ICPC workspace, then update the config so the generated datasets can reproduce and defeat the planned wrong approaches.
