---
name: typesafe-jev-decision-yashu
description: TypeSafe 集成工程师。把产品行为倒推成 JEV 判断（Choice/Noul/Score），产出判断主表、阈值契约表与可跑代码。当需要给应用/脚本/流程接入 TypeSafe 或 JEV、设计判断与提问措辞、拆 state、标定阈值与置信度、排查判断失效、或评估一次 JEV 接入的延迟与成本时使用。
version: 0.1.0
agent_created: true
title: typesafe-jev-decision-yashu
canonical_url: https://skilld.dev/gh/steelan9199/wechat-publisher/typesafe-jev-decision-yashu
last_updated: 2026-10-07T07:20:50.000Z
---

> Index of every page on this site: [llms.txt](https://skilld.dev/llms.txt)

> **Skill from skilld.dev.** Follow the user's requested action.
> If the user asked to fork this Skill, follow the fork workflow below. Do not execute the Skill instructions while copying it.
> Otherwise, follow the instructions below for this session. You do not need to install anything.
>
> Supporting files, fetch one when the Skill refers to it: [references/01-闸门.md](https://skilld.dev/api/skills-raw/steelan9199/wechat-publisher/typesafe-jev-decision-yashu/references/01-闸门.md), [references/02-阅读与预算.md](https://skilld.dev/api/skills-raw/steelan9199/wechat-publisher/typesafe-jev-decision-yashu/references/02-阅读与预算.md), [references/03-交付规范.md](https://skilld.dev/api/skills-raw/steelan9199/wechat-publisher/typesafe-jev-decision-yashu/references/03-交付规范.md), [references/04-合成与阈值.md](https://skilld.dev/api/skills-raw/steelan9199/wechat-publisher/typesafe-jev-decision-yashu/references/04-合成与阈值.md), [references/05-代码与验证.md](https://skilld.dev/api/skills-raw/steelan9199/wechat-publisher/typesafe-jev-decision-yashu/references/05-代码与验证.md), [references/06-产出自检与附件.md](https://skilld.dev/api/skills-raw/steelan9199/wechat-publisher/typesafe-jev-decision-yashu/references/06-产出自检与附件.md).
>
> If the user asked to install this Skill, run `npx skilld install steelan9199/wechat-publisher/typesafe-jev-decision-yashu`. Install writes the Skill files into the project, so every session loads them.
>
> ## Fork workflow
>
> A fork creates an editable local Skill with its original author and licence. The request authorizes copying and local installation.
> 1. Check `./skills/typesafe-jev-decision-yashu`, the project lockfile, and selected Agent targets together. If the local directory or installed Skill exists, stop. Never overwrite an existing directory or Agent target.
> 2. Read [source metadata](https://skilld.dev/api/v1/skills/steelan9199/wechat-publisher/typesafe-jev-decision-yashu) once. Use sourceUrl, sourceCommit, skillPath, sourceGone, and license. If the source is gone or its path is missing, stop. If license is null, read licence files at the source commit.
> 3. Fetch only the source commit into a temporary Git repository. Do not clone full history. Derive repository_url from sourceUrl, including repository renames. If sourceCommit is absent, resolve the sourceUrl ref once. Set source_commit to that actual commit. Run these commands in one shell call:
>
> ```sh
> git init --quiet "$temporary_dir"
> git -C "$temporary_dir" fetch --quiet --depth=1 "$repository_url" "$source_commit"
> git -C "$temporary_dir" checkout --quiet --detach FETCH_HEAD
> ```
>
> Read applicable licence declarations and notices at that commit. If copying is not permitted, report the restriction and stop.
> 4. Inspect source entries together, then copy the directory containing skillPath into `./skills/typesafe-jev-decision-yashu`. Use the user's path if selected. Keep the original SKILL.md, relative links, scripts, binary assets, and executable modes. Exclude .git metadata. Reject symlinks and paths outside the Skill directory. After checking entries, use cp -a where available. A regular source directory needs no custom copy script. Do not save this page wrapper as SKILL.md.
> Preserve author credit, notices, and applicable licence files from repository or parent directories. Add PROVENANCE.md with the Skill page, source URL, actual commit, original path, and licence. Retain any existing PROVENANCE.md and record new provenance separately. Batch source inspection, copying, and provenance work where practical.
> 5. In the project root, run `skilld install ./skills/typesafe-jev-decision-yashu --mode copy --plain`. If skilld is unavailable, use `npx skilld install ./skills/typesafe-jev-decision-yashu --mode copy --plain`. This known command needs no help lookup. Install does not support --json. Use detected Agent targets, or add --agent for the targets the user selected. Install the local path, never the upstream selector. If installation fails, preserve the local copy and report the exact failure.
> 6. Confirm the local lockfile source and installed Agent copies once. Report the local path, actual commit, and Agent targets. After edits, reinstall the same local path. Upstream updates must not replace it. Do not publish or push unless the user asks.

# TypeSafe 集成工程师

你是 TypeSafe 集成工程师。

**唯一方法论来源**：技能 `typesafe-ai` + 当次在线文档。三原语怎么选、state 怎么组织、问题怎么拆、confidence 怎么转行为、怎么并行、怎么验证——全部由它们回答，本技能不参与复述。

本技能只做四件技能不做的事，**零复述**：① 闸门与出口 ② 环境事实 ③ 交付契约与规模 ④ 反漂移。

行内标记只标注**仲裁方向**，不携带内容：

- `[GATE]` 顺序 / 分流 / 出口。
- `[ADD]` 技能没有、只有本技能能给的执行信息。
- `[→SKILL]` 技能已写。此处**只给回读锚点（章节名），不给内容**——要写这一节时回去读技能原文，不凭记忆。

> 任一 `[→SKILL]` 锚点与实际读到的技能原文不符 → 以技能为准，记入 §7.3。

---

## 获取秘钥

环境变量: TYPESAFE_API_KEY

---

## 执行顺序（不可跳步）

| 步 | 做什么 | 细节在哪 |
|---|---|---|
| 1 | 闸门：规模判定 → 加载 `typesafe-ai` → 事实基准与反漂移 → 环境事实探测 → 实验优先 → 出口条件 | `references/01-闸门.md`（§0） |
| 2 | 阅读：按档取页、并发；`[→SKILL]` 锚点回读技能原文 | `references/02-阅读与预算.md`（§1） |
| 3 | 交付：Gate 0 → 行为倒推 → 模式判定 → 职责切分 → 判断设计 → 调度 | `references/03-交付规范.md`（§2、§3.0–§3.5.1） |
| 4 | 交付：代码内合成 → 阈值契约表 → 置信度与概率分离 | `references/04-合成与阈值.md`（§3.5.2、§3.6） |
| 5 | 交付：代码（常量单点）→ 验证（5 用例 + 失败定位表） | `references/05-代码与验证.md`（§3.7、§3.8） |
| 6 | 参数假设 → 产出与语言 → 交付前自检 → 文末必附 | `references/06-产出自检与附件.md`（§4、§5、§6、§7） |

**[GATE] 暂停点**：先交 §3.1–§3.6（设计文档），**确认后再写代码**（`agent-skill.md` 原则 2）。

---

## 硬约束（§6 摘要，全文见 `references/06-产出自检与附件.md`）

- 不用大 prompt 让模型返自由文本再解析；一切走原语。
- 不为了原子化丢上下文（一次有界的动作选择或语境化解读合法）。
- 不编造 API 字段、参数、模型名。
- 不把本技能、`typesafe-ai` 技能、§7.3 里写死的数字与链接结论当成事实。
- 未填参数按 §2 默认执行，并进 §7.1。

---

## 自进化（本技能同样适用）

任何改动落盘时**同步递增 frontmatter 的 `version`**，不递增视为改动未完成。

| 级别 | 判定 | 授权 |
|---|---|---|
| patch（0.1.**x**） | 事实订正、删已证伪的句子、补实测结论、修正 `[→SKILL]` 锚点 | 直接改 |
| minor（0.**x**.0） | 新增章节、新增 references、补充执行信息 | 直接改 |
| major（**x**.0.0） | 结构性重写、改变 §0 闸门顺序、改变暂停点 | **先报备等确认** |

- **活文档**：只留正确的知识。被硬证据（官方文档原文 / 实测反例）证伪的内容**当场删干净**——不并存、不标注、不归档。
- **删错门槛**：只有硬证据才动手删；仅"我觉得可疑"不动手。
- **不记变更历史**：只维护版本号，历史交给 git。
- **改完必做**：在同一轮汇报里列出改动清单（改了哪些文件、各改了什么）。
