القائمة
运行、长测、诊断和优化 Windows 扫雷自动游玩流程。用于用户要求自动打开扫雷、自动玩、跑单局或多局测试、监控长测、分析失败截图、排查识图/几何/点击/策略问题、或继续迭代 `apps/minesweeper_autoplay/workflows/minesweeper_solver.py` 时。
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
| name | minesweeper-autoplay |
| description | 运行、长测、诊断和优化 Windows 扫雷自动游玩流程。用于用户要求自动打开扫雷、自动玩、跑单局或多局测试、监控长测、分析失败截图、排查识图/几何/点击/策略问题、或继续迭代 `apps/minesweeper_autoplay/workflows/minesweeper_solver.py` 时。 |
../../../apps/minesweeper_autoplay/../../../apps/minesweeper_autoplay/workflows/minesweeper_solver.py../../../runtime/apps/minesweeper_autoplay/test_artifacts/../../../runtime/knowledge_runs/../../../apps/minesweeper_autoplay/scripts/run-solver.ps1../../../skills/task_skills/minesweeper_autoplay/README.mdsingle: 单局跑到自然胜负或单局结束即停retry: 最多跑 N 局until_success: 持续重试直到成功，或直到额外限制命中apps/minesweeper_autoplay/scripts/run-solver.ps1 unless there is a good reason to call the Python entry directly.
../../../apps/minesweeper_autoplay/workflows/minesweeper_solver.py, run py_compile, then rerun a targeted smoke test before longer tests.Use the wrapper script for normal operation:
powershell -ExecutionPolicy Bypass -File apps/minesweeper_autoplay/scripts/run-solver.ps1 -Mode single
Read references/run-modes.md for the supported parameter combinations.
Read references/usage.md for common user-facing invocation patterns.
../../../runtime/apps/minesweeper_autoplay/test_artifacts/solver_stop_summary.txt when present.attempt_*_step_*_consensus_01.png and terminal failure screenshots.cell_action_diagnostics.txt for no_effect, blocked, wrong_target, or visual-sync clues.Read references/diagnostics.md and references/artifacts.md when diagnosing non-trivial failures.
platform/src/omniauto/skills/.../../../skills/ aligned with runtime bundle updates in .agents/skills/.references/run-modes.mdreferences/usage.mdreferences/diagnostics.mdreferences/artifacts.md运行、诊断、补跑并生成 1688 关键词调研报告。用于按关键词抓取前 N 页、使用价格排序搜索、抽样详情页、生成完整 HTML 报告、在验证码出现时切到人工接管、以及复盘 1688 调研产物时。
指导 Kimi 将复杂浏览器自动化任务拆分为确定性 Workflow + AtomicStep 执行， 禁止在运行时用 AI 做视觉决策或实时思考。适用于电商抓取、数据调研、报告生成等多步骤任务。 触发条件：用户要求执行需要浏览器操作的复杂任务（如"抓取XX前N页并生成报告"、"调研某网站数据"）。
Use when working on OmniAuto knowledge closeout, observation emission, manual closeout, or AI-assisted candidate generation. This skill governs both AI behavior and the platform policy boundaries for automatic knowledge growth, and should be used whenever a task touches `platform/src/omniauto/knowledge/`, `knowledge/`, `omni closeout`, or candidate promotion rules.