with one click
autojudge-contest
autojudge-contest contains 5 collected skills from werlang, with repository-level occupation coverage and site-owned skill detail pages.
Skills in this repository
Use when changing AutoJudge Contest VS Code extension runtime behavior, including login/logout flows, session restore, contest tree rendering, problem and submission actions, submission workflows, output-channel rendering, or AutoJudge API error handling.
Use when changing the AutoJudge Contest VS Code extension manifest surface, including views, commands, menus, settings, context keys, or user-facing documentation that must stay synchronized with package.json.
Use when a task needs repository commands or local tooling so the work runs through the Compose development container instead of the host machine.
Use when refactoring or changing behavior in this repository so work follows a Vitest-first workflow executed inside the Compose development container.
Human-like long-term memory across sessions. Run memory_search at task start unless the user opts out, run additional searches only on mid-task triggers, persist reusable outcomes with memory_write, and prune stale or duplicate entries with memory_forget so active memory stays small and relevant.