菜单
Step-by-step workflow to identify fallback-driven regressions and replace them with strict source-level validation and fail-closed transitions.
默认使用会先检查来源的 Prompt；你也可以切换为直接命令，或下载本地副本。
决定是否安装前，请先阅读 SKILL.md，以及 SkillsMP 当前展示的配套文件。
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
直接命令不会经过审查 Prompt；运行前请先检查来源。
npx skills add https://github.com/MRiabov/Problemologist-AI --skill source-fix-remediation-workflow命令会保持在同一行。复制前请横向滚动并检查完整内容。
想先保存到本地？可下载 SkillsMP 当前能够提供的文件。
Benchmark planning and handoff authoring for Problemologist. Use when creating or revising benchmark planner artifacts (`benchmark_plan.md`, `todo.md`, `benchmark_definition.yaml`, `benchmark_assembly_definition.yaml`, `benchmark_plan_evidence_script.py`, `benchmark_plan_technical_drawing_script.py`), checking benchmark solvability or randomization, defining benchmark-owned fixture motion, reviewing planner drafting output with `render_technical_drawing()`, enforcing exact-grounded inventory mentions, preparing the plan for `submit_benchmark_plan()`, or inspecting simulation evidence through frame-indexed `objects.parquet` sidecars.
Benchmark execution-review skill for validating implemented benchmarks after validation and simulation. Use when reviewing `benchmark_script.py`, `benchmark_definition.yaml`, `benchmark_assembly_definition.yaml`, `validation_results.json`, `simulation_result.json`, `scene.json`, render or video evidence, frame-indexed `objects.parquet` sidecars, or stage-specific review artifacts; when applying the benchmark execution review checklist for exact inventory grounding, geometry validity, solvability, randomization, and benchmark-side motion; or when writing the `reviews/benchmark-execution-review-*.yaml` pair and submitting via `bash scripts/submit_review.sh`.
Benchmark implementation role for turning approved benchmark plans into benchmark_script.py and helper modules, validating and simulating benchmark revisions, preserving planner inventory exactness, handling benchmark-side motion or COTS-backed fixture geometry, and refusing only when the approved plan is infeasible. Use when working as the Benchmark Coder after plan approval, when fixing benchmark validation or simulation failures, when producing review-ready benchmark evidence, when deciding whether to write plan_refusal.md, or when inspecting simulation evidence through frame-indexed `objects.parquet` sidecars.
基于 SOC 职业分类
正在显示 SKILL.md
| name | source-fix-remediation-workflow |
| description | Step-by-step workflow to identify fallback-driven regressions and replace them with strict source-level validation and fail-closed transitions. |
Use this skill to execute the fix (not just policy) when fallback behavior causes invalid outputs to pass.
./scripts/run_integration_tests.sh for integration behavior.dataset/evals/run_evals.py with explicit agent/task scope.PLANNED, COMPLETED, etc.).fallback, default, heuristic string checks).benchmark_plan.md / engineering_plan.md / todo.md / YAML for recovery.