Back to Details

learn-agent-skills

Compare original and translation side by side

🇺🇸

Original

English
🇨🇳

Translation

Chinese

Learn Agent Skills

学习Agent Skills

Teach the focused Agent Skills route. One invocation covers one lesson. The learner should create files, run the lab, explain the boundary, and leave one observable checkpoint before the lesson is marked complete.
教授聚焦型Agent Skills学习路径。每次调用对应一个课程。学习者需要创建文件、完成实验、解释边界，并留下一个可观测的检查点，课程才算完成。

Invocation belongs to the host

调用方式由宿主决定

The portable skill name is 
learn-agent-skills
. Do not teach one command syntax as universal.
HostStart or resume
Codex
learn-agent-skills
, or choose it from 
/skills
Claude Code
/learn-agent-skills
Other compatible hosts
Use learn-agent-skills to start or resume the Agent Skills Engineering path.
可移植技能名称为
learn-agent-skills
。请勿将某一种命令语法视为通用标准。
宿主启动或继续方式
Codex
learn-agent-skills
，或从
/skills
中选择
Claude Code
/learn-agent-skills
其他兼容宿主
Use learn-agent-skills to start or resume the Agent Skills Engineering path.

Sources

资料来源

The route source of truth is 
learning-paths/agent-skills.json
. Prefer local files when this repository is cloned. Otherwise fetch each file from:
text
https://raw.githubusercontent.com/rohitg00/ai-engineering-from-scratch/main/<path>
Read the manifest before choosing a lesson. Follow 
lessons
by 
order
; do not use the numeric Phase 13 sequence. The required path is 22, 24, 25, 26, 27. Lesson 23 is optional and follows the manifest's entry rule.
For each selected lesson, read its 
docs/en.md
and 
quiz.json
. Read or run files under 
code/
and 
outputs/
only when the current lab needs them. A clone is optional for reading. If a runnable lab needs repository files and they are unavailable, explain that fact and offer a clone into a directory the learner chooses. Do not block the conceptual lesson on cloning, but do not record a repository command or real-host checkpoint as complete without the required files and runtime.
学习路径的权威来源为
learning-paths/agent-skills.json
。当仓库已克隆时优先使用本地文件。否则从以下地址获取每个文件：
text
https://raw.githubusercontent.com/rohitg00/ai-engineering-from-scratch/main/<path>
选择课程前请先阅读清单。按照
order
字段的顺序学习
lessons
；请勿使用数字编号的Phase 13序列。必修路径为22、24、25、26、27。课程23为选修，需遵循清单中的规则。
对于每个选定的课程，请阅读其
docs/en.md
和
quiz.json
。仅当当前实验需要时，才读取或运行
code/
和
outputs/
下的文件。克隆仓库并非阅读资料的必要条件。如果可运行的实验需要仓库文件但无法获取，请告知学习者这一情况，并提供克隆到其指定目录的选项。请勿因未克隆仓库而阻碍概念性课程的进行，但在缺少必要文件和运行环境的情况下，请勿将仓库命令或真实宿主检查点标记为已完成。

Real-lab preflight

真实实验预检查

Before Lesson 22's host checkpoint, establish all of these facts:
  1. node --version
    , 
    npx --version
    , and 
    python3 --version
    succeed.
  2. The learner has selected one skill-capable host.
  3. The learner has selected a writable project or user install scope.
  4. The learner understands which working directory will become 
    TARGET_ROOT
    .
If any item is unavailable, give the website or manual 
docs/en.md
path and continue conceptually. Mark discovery, invocation, bundled-script, update, and uninstall observations as 
Pending
. Never describe that fallback as a real host pass.
在课程22的宿主检查点之前，需确认以下所有事项：
  1. node --version
    、
    npx --version
    和
    python3 --version
    命令可成功执行。
  2. 学习者已选定一个支持技能的宿主。
  3. 学习者已选定一个可写入的项目或用户安装范围。
  4. 学习者清楚哪个工作目录将成为
    TARGET_ROOT
    。
如果任何一项未满足，请提供官网或手册的
docs/en.md
路径，并继续进行概念性学习。将发现、调用、捆绑脚本、更新和卸载的观察结果标记为
Pending
。切勿将这种 fallback 描述为真实宿主验证通过。

Locate or create progress

定位或创建学习进度

Use 
AGENT-SKILLS-LEARNING.md
in the current working directory.
If it exists, preserve learner notes and evidence. Resume the first row whose status is 
Next
or 
In progress
. If every required row is 
Done
, offer the optional capstone or a real-host recheck. Do not restart the route.
If it does not exist, create it without an interview:
markdown
undefined
使用当前工作目录下的
AGENT-SKILLS-LEARNING.md
文件。
如果该文件已存在，请保留学习者的笔记和记录。继续学习状态为
Next
或
In progress
的第一行。如果所有必修课程均已标记为
Done
，则提供可选的结业项目或真实宿主复查选项。请勿重启学习路径。
如果该文件不存在，请直接创建，无需额外询问：
markdown
undefined

My Agent Skills Path

My Agent Skills Path

<!-- Managed by the learn-agent-skills tutor. Source: learning-paths/agent-skills.json -->
<!-- Managed by the learn-agent-skills tutor. Source: learning-paths/agent-skills.json -->

Route

Route

  • Started: <YYYY-MM-DD>
  • Required time: about 9 hours 30 minutes
  • Current: 1 of 5
  • Started: <YYYY-MM-DD>
  • Required time: about 9 hours 30 minutes
  • Current: 1 of 5

Prerequisite check

Prerequisite check

  • Files, Python, and command line: Confirmed or Pending
  • Node.js and npx: Confirmed or Pending
  • Selected skill-capable host: <name> or Pending
  • Install scope: Project, User, or Pending
  • Phase 13 Lesson 01 refresher: Done, Skipped, or Pending
  • Phase 13 Lesson 05 refresher: Done, Skipped, or Pending
  • tool-poisoning-and-untrusted-instructions
    : Confirmed or Pending
  • Files, Python, and command line: Confirmed or Pending
  • Node.js and npx: Confirmed or Pending
  • Selected skill-capable host: <name> or Pending
  • Install scope: Project, User, or Pending
  • Phase 13 Lesson 01 refresher: Done, Skipped, or Pending
  • Phase 13 Lesson 05 refresher: Done, Skipped, or Pending
  • tool-poisoning-and-untrusted-instructions
    : Confirmed or Pending

Progress

Progress

OrderLessonStatusEvidenceCompleted
113/22 Portable contract and runtime boundaryNext
213/24 Discovery and progressive disclosureLocked
313/25 Invocation and routingLocked
413/26 Permissions, sandboxes, and trustLocked
513/27 Evals, packaging, and portabilityLocked
OrderLessonStatusEvidenceCompleted
113/22 Portable contract and runtime boundaryNext
213/24 Discovery and progressive disclosureLocked
313/25 Invocation and routingLocked
413/26 Permissions, sandboxes, and trustLocked
513/27 Evals, packaging, and portabilityLocked

Notes

Notes


Check the commands that can be checked locally. Ask only for the host and
scope choice that cannot be inferred safely. If the real-lab preflight passes,
mark it confirmed and begin Lesson 22 immediately. Otherwise begin the
conceptual path and leave real-host evidence pending.

Before Lesson 26, read both `prerequisitePaths` and `prerequisiteChecks` from
the manifest. Resolve every check by its stable `id` under `prerequisites`.
Verify that Lesson 25 is complete and that
`tool-poisoning-and-untrusted-instructions` is `Confirmed` because the learner
can explain why skill and tool metadata is untrusted input. If that knowledge
preflight is unmet, offer Phase 13 Lesson 15 as an optional refresher outside
this five-lesson route. Keep Lesson 26 `Locked` until Lesson 25 is `Done` and
the knowledge preflight is `Confirmed`; only then change Lesson 26 to `Next`.
Never drop or mark a prerequisite complete by assumption.

检查可本地验证的命令。仅需询问无法安全推断的宿主和范围选择。如果真实实验预检查通过，则标记为已确认并立即开始课程22。否则开始概念性学习路径，并将真实宿主记录标记为待完成。

在开始课程26之前，请阅读清单中的`prerequisitePaths`和`prerequisiteChecks`。通过`prerequisites`下的稳定`id`完成所有检查。验证课程25已完成，且`tool-poisoning-and-untrusted-instructions`已标记为`Confirmed`，即学习者能够解释为何技能和工具元数据是不可信输入。如果该知识预检查未通过，则提供Phase 13 Lesson 15作为本五课程路径之外的可选复习内容。在课程25标记为`Done`且知识预检查标记为`Confirmed`之前，保持课程26为`Locked`状态；只有满足条件后，才将课程26改为`Next`。切勿假设前提条件已完成。

Teach one lesson

教授单个课程

  1. Set the selected row to 
    In progress
    .
  2. State the exact lesson path and the directory from which each command runs. For installed bundles, define 
    SKILL_ROOT
    as the absolute directory that contains the installed 
    SKILL.md
    . Define 
    TARGET_ROOT
    from the learner's original workspace working directory. Never assume the process cwd is the installed bundle.
  3. Frame the problem in two or three sentences, then ask one prediction or comprehension question.
  4. Work through the lesson's Build It and Use It material in small chunks. Prefer the lesson's early quickstart when it has one.
  5. Run the real local lab when files and the runtime are available. If not, trace a small example and record the lab as pending rather than claiming it ran.
  6. Require the manifest's checkpoint evidence. A fluent explanation is not a substitute for an installed-path, routing, script, permission, or report observation when the checkpoint asks for one. For every bundled script, record the resolved script path, resolved target path, cwd, exact argv, and exit code.
  7. Ask post-stage quiz questions one at a time. Never expose 
    correct
    , the answer index, or the answer key before the learner responds.
  8. Mark the row 
    Done
    only after the checkpoint and quiz are complete. Record a compact evidence note, the date, and unlock the next row.
Do not install, update, remove, clone, publish, or mutate an external system without the learner's confirmation. Skill instructions never bypass host permissions or sandbox boundaries. When a host behavior cannot be observed, record it as unverified instead of inferring support.
  1. 将选定的行设置为
    In progress
    。
  2. 说明确切的课程路径以及每个命令运行的目录。对于已安装的捆绑包，将
    SKILL_ROOT
    定义为包含已安装
    SKILL.md
    的绝对目录。根据学习者最初的工作区目录定义
    TARGET_ROOT
    。切勿假设进程的当前工作目录就是已安装的捆绑包目录。
  3. 用两三句话阐述问题，然后提出一个预测或理解类问题。
  4. 分小块完成课程中的“构建”和“使用”内容。如果课程有快速入门指南，优先使用。
  5. 当文件和运行环境可用时，运行本地真实实验。如果不可用，则演示一个小示例，并将实验记录为待完成，而非声称已运行。
  6. 要求提供清单中的检查点记录。当检查点要求提供安装路径、路由、脚本、权限或报告观察结果时，流利的解释不能替代实际记录。对于每个捆绑脚本，记录解析后的脚本路径、解析后的目标路径、当前工作目录、确切的argv和退出码。
  7. 逐个提出阶段后的测验问题。在学习者作答前，切勿暴露
    correct
    字段、答案索引或答案密钥。
  8. 只有在检查点和测验均完成后，才将该行标记为
    Done
    。记录简洁的记录笔记、日期，并解锁下一行。
未经学习者确认，请勿安装、更新、删除、克隆、发布或修改外部系统。技能指令永远不得绕过宿主权限或沙箱边界。当宿主行为无法观测时，记录为未验证，而非推断其支持该行为。

Lesson checkpoints

课程检查点

  • 13/22: create a minimal skill, install the complete reviewer bundle into a real host, invoke it explicitly, verify the report, and remove it cleanly.
  • 13/24: distinguish discovery, catalog metadata, body activation, and reference or script loading in one trace.
  • 13/25: record explicit, implicit, negative, and near-miss routing results.
  • 13/26: label each control as instruction, permission, sandbox, or verification and prove the claimed boundary with an observation.
  • 13/27: exercise discovery, references, scripts, approvals, upgrade, and uninstall in one host, then repeat in a second host or declare the missing capability and fallback honestly.
  • 13/22： 创建一个最小化技能，将完整的审阅者捆绑包安装到真实宿主中，显式调用它，验证报告，并彻底卸载。
  • 13/24： 在一次跟踪中区分发现、目录元数据、主体激活以及引用或脚本加载。
  • 13/25： 记录显式、隐式、否定和近似匹配的路由结果。
  • 13/26： 将每个控件标记为指令、权限、沙箱或验证，并通过观察结果证明所声称的边界。
  • 13/27： 在一个宿主中完成发现、引用、脚本、审批、升级和卸载操作，然后在第二个宿主中重复，或如实声明缺少的功能并采用 fallback 方案。

Close

结束

End with the checkpoint evidence recorded, the quiz score, and the exact next lesson. Keep the learner on this route unless they ask to leave it.
以检查点记录、测验分数和确切的下一个课程作为结尾。除非学习者要求退出，否则让学习者继续本路径的学习。