给 Grok Bot 接入 TypeSafe Jev 路由 skill：先 shadow 一周再放行，保留 kill switch

作者教程：用 6 段可复制指令让 Grok Bot 装 TypeSafe SDK、写 router 模块和 skill，先 shadow 模式只记日志一周，再切 active，并保留两个 kill switch

给谁

已在用 Grok Bot 跑例行任务、常遇到重复重试或重复调研，想加一层「下一步该不该做」判断的进阶用户（需会看 Python 和日志）

你会得到

得到一个 jev-usage-router skill：在开浏览器、调研、重试和再开 Bot 前先问 Jev，五选一返回 proceed / reuse_cache / stop_retry / research_capped / ask_human，并把每次决策写进 decisions.jsonl。作者未给出量化节省数字，要求读者用自己的流量对比 shadow 周与 active 周

你需要

Grok Bot · TypeSafe Jev（typesafe-sdk）

步骤

1. 建 API key

   在 TypeSafe 控制台创建 API key。作者要求：不要放进聊天、仓库或任何你打算分享的 skill。

2. 用掩码流程把 key 放上云电脑

   把下面这段发给 Grok Bot，走掩码密钥请求流程，key 不经过聊天。注意（原文）：每个账号只有一台云电脑，你跑的每个 Bot 都能看到同样的文件、会话和登录，放上去的 key 对所有 Bot 可读。

   搭建指导 · 来源原文。交给协助搭建的 Agent；不是节点运行指令。

   复制提示词

   I need to give you an API credential for TypeSafe. Use the masked secret
   request flow, not chat. The value must land as the environment variable
   TYPESAFE_API_KEY on your computer, and it must not appear in any file,
   log, skill or repo you create. Confirm back only that the variable is set
   and readable, never its value.

3. 装 SDK 并冒烟测试

   让 Bot 安装 typesafe-sdk 并写 /work/jev/smoke.py 发一次 Choice。输出里要看到三样：你自己列表中的一个值、概率分布、一个置信度数字；缺任何一样，说明 SDK 连的不是它。

   搭建指导 · 来源原文。交给协助搭建的 Agent；不是节点运行指令。

   复制提示词

   On your computer, run:  pip install typesafe-sdk
   
   Then write /work/jev/smoke.py that opens TypeSafeClient(model="jev-latest")
   and sends one system_one call with a single Choice question. State:
   "goal: file a support ticket. task: routing. last error: none x0".
   Criteria: billing, technical, sales.
   
   Run it and show me the raw response object: the choice, the probabilities
   and the confidence. If TYPESAFE_API_KEY is missing, stop and tell me,
   do not hardcode anything.

4. 写 router 模块

   让 Bot 按规格写 /work/jev/router.py：只把 goal、task_type、last_error、repeats 四个字段作为 state 发给 Jev（不发提示词、仓库或对话），五个选项写死，每次调用追加到 decisions.jsonl，行为读 config.json，默认关闭；运行前先给你看文件。原文附 Python 调用片段可对照。

   搭建指导 · 来源原文。交给协助搭建的 Agent；不是节点运行指令。

   复制提示词

   Build a router module at /work/jev/router.py with one function:
   
       route(goal: str, task_type: str, last_error: str, repeats: int) -> dict
   
   It sends exactly those four fields to Jev as the state, with one Choice
   question named "action" and these criteria:
   
     proceed          nothing suggests the next step will fail
     reuse_cache      a fresh artifact for this already exists
     stop_retry       the same approach failed the same way
     research_capped  enough sources are already gathered
     ask_human        the step sends, publishes, pays or deletes
   
   Return the choice, the confidence and the probabilities. Append every call
   to /work/jev/decisions.jsonl with a timestamp, the four inputs and the full
   response. Read behaviour from /work/jev/config.json with two keys:
   "enabled" and "mode" ("shadow" or "active"). Default to enabled false.
   
   No prompt text in the state beyond those four fields. Show me the file
   before you run it.

   搭建指导 · 来源原文。交给协助搭建的 Agent；不是节点运行指令。

   复制提示词

   from typesafe_sdk import Choice, TypeSafeClient
   
   with TypeSafeClient(model="jev-latest") as client:   # reads TYPESAFE_API_KEY
       r = client.system_one(
           state=f"goal: {goal}\ntask: {task_type}\nlast error: {err} x{repeats}",
           questions={
               "action": Choice(
                   instructions="What should the agent do next",
                   criteria={
                       "proceed": "Nothing suggests the next step will fail",
                       "reuse_cache": "A fresh artifact for this already exists",
                       "stop_retry": "The same approach failed the same way",
                       "research_capped": "Enough sources are already gathered",
                       "ask_human": "The step sends, publishes, pays or deletes",
                   },
               )
           },
       )
   
   print(r.choices["action"].choice, r.choices["action"].confidence)

5. 建 skill「jev-usage-router」

   触发点：任何浏览器会话、调研、失败重试之前，以及再开一个 Bot 之前；按返回的五种动作执行，ask_human 无论置信度多高都不能被覆盖。作者提醒：skill 的权威副本放在你自己的版本库，Grok Bot 每个 routine 只保留最近 20 条运行记录。

   运行提示词 · 来源原文。使用前确认粘贴位置、输入变量及输出格式。

   复制提示词

   Create a skill called jev-usage-router.
   
   Trigger: before any browser session, any research pass, any retry of a
   failed step, and before spawning an extra bot.
   
   Steps: call route() from /work/jev/router.py with the current goal, the
   task type, the text of the last error and how many times it repeated.
   Then honour the action:
   
     proceed          continue as planned
     reuse_cache      use the existing artifact, skip the work
     stop_retry       abandon this approach and report why
     research_capped  stop at 5 sources and write up what you have
     ask_human        pause and ask me, do not proceed on your own
   
   Approval boundary: ask_human is never overridden, whatever the confidence.
   Output: one line per decision in the run report.

6. shadow 模式跑一周

   设 {"enabled": true, "mode": "shadow"}：每个触发点都调用并记日志，但照原样做事；每天把调用数、动作分布、平均置信度和分歧写入 report.md。作者理由：便宜的判断器也会错，例如 stop_retry 可能拦掉第四次本可成功的尝试；少于一周样本太小。

   运行提示词 · 来源原文。使用前确认粘贴位置、输入变量及输出格式。

   复制提示词

   Set /work/jev/config.json to {"enabled": true, "mode": "shadow"}.
   
   In shadow mode, call the router at every trigger and log the decision, then
   ignore it and do what you would have done anyway. Change nothing about how
   you work.
   
   At the end of each day, append to /work/jev/report.md: how many calls,
   the action split, the mean confidence, and every case where the logged
   action disagreed with what you actually did.

7. 切 active，保留 kill switch

   读完日志再切 active：开始执行决策，但 ask_human 仍暂停等你；两个 kill switch 始终可用（任一消息里写「bypass jev」，或 config 设 enabled:false），决策日志照旧写。作者建议先只路由那一个总在同样失败上重试的 routine。

   运行提示词 · 来源原文。使用前确认粘贴位置、输入变量及输出格式。

   复制提示词

   Set mode to "active". From now on, honour the action instead of logging it.
   
   Two kill switches stay available to me at all times:
     1. "bypass jev" in any message, which ignores the router for that run
     2. {"enabled": false} in config.json, which disables it entirely
   
   ask_human still pauses for me. Keep writing decisions.jsonl exactly as before.

Create a skill called jev-usage-router.

Trigger: before any browser session, any research pass, any retry of a
failed step, and before spawning an extra bot.

Steps: call route() from /work/jev/router.py with the current goal, the
task type, the text of the last error and how many times it repeated.
Then honour the action:

  proceed          continue as planned
  reuse_cache      use the existing artifact, skip the work
  stop_retry       abandon this approach and report why
  research_capped  stop at 5 sources and write up what you have
  ask_human        pause and ask me, do not proceed on your own

Approval boundary: ask_human is never overridden, whatever the confidence.
Output: one line per decision in the run report.