此技能识别 GitHub Agentic Workflows 相关请求的类型,并引导代理读取对应提示或技能文件,涵盖设计、创建、配置、更新、调试和升级。
- 设计或创建新工作流
- 更新现有工作流
- 调查工作流错误
这是供 `cli-exercise` 内部使用的准备检查模块,用于核验构建、终端和媒体工具是否可用;缺少依赖时先征得安装同意,安装后重新检查。
Internal readiness and installation-consent flow for cli-exercise. Requires entry-skill context and does not independently authorize a recording or installation.
实测评分 未实测
查看安装方法这是一个内部模块,用于检查运行 cli-exercise 所需的工具是否就绪。它通过 Go 辅助程序检查 Node/Tuistory 接口和伪终端、ffmpeg/ffprobe 与指定编码器、Go 字体渲染能力,以及用例所需的 Shell 或可执行文件。如果依赖缺失,会先检查是否已有合适的安装,再列出拟采用的版本、来源、目标位置和命令。安装前必须获得同意,完成后会再次检查相关能力。
在使用 cli-exercise 开始录制前,可用此模块检查环境是否准备妥当。它可以确认会话视频所需的 MP4 输出能力,也可以检查请求中指定的 GIF 输出。检查结果有助于了解当前可用能力、缺少的依赖,以及是否能继续准备。
此模块不能单独调用,只能在 cli-exercise 上下文中作为内部参考使用。安装依赖需要请求者批准。此类批准不代表允许录制、修改文件、向 GitHub 写入或执行其他操作。检查结果仅说明环境准备情况,不会授予权限,也不会提供操作系统隔离。
本介绍由 AI(OpenAI Codex)根据公开的 SKILL.md 撰写。准确的规格请查看下方原文。
AI 根据 SKILL.md 的内容检查了需要注意的操作和使用前提。评分只依据实际运行的结果(实测对比)。
不计入评分。 由 AI(OpenAI Codex)于 2026年10月26日 评估。仅供参考,安装前请同时查看原文。
安装前可以查看交给 AI 代理的指令内容。
Use the package root containing the user-facing cli-exercise/SKILL.md.
This module is bundled with that skill and is loaded as a referenced resource.
If the caller supplies a ready receipt already checked for this task, read and reuse it when its selected tools and probed formats match the case. Do not repeat probes or propose installation just because the receipt was produced elsewhere in the same task. Recheck if its current applicability is uncertain. A receipt does not grant installation or test-resource authority. Normal session delivery, even for one case, needs MP4 chapter intermediates; confirm MP4 capability as well as any requested GIF output.
When a check is needed, first follow the offline-first Go build instructions in the prerequisite reference. Build the helper once for this task, outside the skill and repository, then use that executable:
<helper> --skill-root <package-root> preflight check --output <workspace>/preflight.json
Keep Go toolchain auto-downloads and module downloads disabled on the first build. If a locked dependency is missing from the selected cache, show the packages, sources, destination and build command, then ask before enabling its download. A compile error is not permission to try an online build.
Explicit existing Node module, executable, and font paths can be supplied using the documented overrides. Never embed operator-specific paths in the skill.
Check usable capabilities: the Node/Tuistory API and native PTY, ffmpeg/ffprobe and requested encoders, and actual Go font rasterization. Also identify any case-specific shell, editor, or executable. A command on PATH does not prove that its native bindings or required API work.
When a dependency is missing or incompatible:
Use an appropriate dependency location outside the skill and repository. Do not globally update tools or change project dependency manifests by default. If system installation or authentication is required, state that separately. Never enter an OS password or turn a failed privileged command into success.
If approval is declined, unavailable, or the install/recheck fails, report
blocked with the precise remedy. Preserve diagnostics and any partial-install
facts. Do not silently change recorders, output formats, or supplied cases.
The helper only checks capabilities. Perform approved setup with the appropriate package manager.
Dependency approval does not authorize GitHub writes, filesystem changes outside the workspace, credential changes, purchases, or other effects. Obtain the required case-specific authorization separately and record its actual source.
The resulting receipt identifies tools used by the runtime and renderer. It is readiness evidence, not an OS sandbox or a permission bypass.
还没有评价。欢迎分享你的使用感受。
无需登录即可发表。欢迎分享使用感受、提问,或说明在哪些环境下无法运行。
还没有评论。来写第一条吧。
此技能识别 GitHub Agentic Workflows 相关请求的类型,并引导代理读取对应提示或技能文件,涵盖设计、创建、配置、更新、调试和升级。
Exercise a real command-line program with specified cases or a stated goal, then capture and verify the observed behavior. It produces local media evidence such as a session video, helping reviewers see what the command did without treating the recording as proof of general correctness.
中文简介正在准备中。当前显示英语简介。
将命令行练习的预期与记录证据逐项核对,按需渲染并检查媒体,返回各案例结果和本地成果文件路径;默认不发布证据。
此技能按原样执行指定的命令行测试,保留命令、输入、顺序和预期条件;逐项准备、操作并验证,再依据实际证据标记通过、失败、受阻或仅作观察。
此技能根据目标和已批准的范围选择可观察的命令行检查,逐项完成准备、操作和即时验证,并整理证据、尚未达到的行为及检查结果。
通过共享的 Go 运行时和 Tuistory 适配器执行并记录 CLI 案例,按获批流程完成准备、操作、验证和清理,保留终端表现、执行顺序及失败时的部分证据。