What problem does it solve? Maintaining the Optimization Compass repository requires editing canonical knowledge sources without corrupting generated artifacts, breaking stable entity IDs, or silently changing dataset releases. This Skill enforces the correct authority files, evidence rules, and validation tiers for every change type. ## Core Features & Use Cases - Task Classification: Routes every request into one of eight classes (prose correction, content addition, Gallery case, comparison, problem instance, canonical entity, visualization scenario, or release/publishing) with a matching recipe. - Invariant Enforcement: Blocks hand-editing of generated outputs such as knowledge.sqlite, site/public/data, and release files, and requires authoritative sources for factual claims. - Tiered Validation: Prescribes Tier A/B/C validation commands (pytest, ruff, mypy, site builds, Playwright E2E) matched to change risk. - Use Case: When adding a new optimization method article or publishing a dataset release to GitHub Pages, follow the recipes to edit only canonical inputs, regenerate projections, and verify deployment identity with the pages checkpoint script. ## Quick Start Ask the assistant to add a new Gallery case to the Optimization Compass repository following the maintenance skill's recipe and validation tier.
