What problem does it solve? When working on a long-running game project, it is hard to know which problems are genuinely open versus already solved, what evidence a claim needs before it can be stated, and how to avoid overselling results. This Skill provides a curated map of the kami-kakushi project's research frontier so agents and contributors pick ambitious work with clear falsifiable success criteria. ## Core Features & Use Cases - Frontier problem catalog: Documents open game problems (fun certification, CSS-only art direction under Andon Steel, authored narrative depth at T1+) with why generic approaches fail, existing in-repo assets, first concrete steps, and falsifiable result criteria. - Process asset registry: Inventories unusual working systems (shared-tree multi-agent protocol, 21-gate verify lattice, sim-priced design, narrative-gen pipeline, capture-to-drain loop) with pointers to their owning skills and evidence artifacts. - Evidence bar and no-oversell rules: Enforces a four-part methodology (predict numbers first, one mechanism explains all observations, adversarial refutation, checks must be able to fail) plus positioning rules that forbid unearned novelty or SOTA claims. - Use Case: When a session is about to claim something is novel or solved, or when choosing the next big bet, load this Skill to check the claim against the evidence bar and route the idea through brainstorm, plan, build, and verdict stages. ## Quick Start Ask the agent to load the kami-research-frontier skill and evaluate whether a proposed idea is a genuine frontier bet or routine work before writing it up.
