niceeval 0.12.1 → 0.12.2-canary.31

package/docs-site/zh/examples/integrations/pi-sdk.mdx

@@ -11,14 +11,16 @@ description: "一个 pi-agent-core(@earendil-works)助手后端，接入 NiceEva
 - **before**：[https://github.com/CorrectRoadH/niceeval/tree/main/examples/zh/origin/pi-sdk](https://github.com/CorrectRoadH/niceeval/tree/main/examples/zh/origin/pi-sdk) —— 独立的 `@earendil-works/pi-agent-core` HTTP 服务，还没接任何评估用例。
 - **after**：[https://github.com/CorrectRoadH/niceeval/tree/main/examples/zh/tier1/pi-sdk](https://github.com/CorrectRoadH/niceeval/tree/main/examples/zh/tier1/pi-sdk) —— 同一个应用接入 NiceEval 之后的样子。
 
-**接入方式**：官方转换器——pi 原生 `AgentEvent` → 标准事件的映射是
-`createPiAgentEventStream`（`"niceeval/adapter"` 导出）的事，adapter 只剩传输粘合：
-应用在哪个 URL、审批打哪个端点（`calculate` 工具走 HITL 审批）。应用由你自己按它的
-方式启动（`pnpm start`），评估用例不代管进程。应用侧 `src/backend/*` 逐字节未变。
+**接入方式**：`createPiAgentEventStream`（从 `"niceeval/adapter"` 导出）把 pi 原生 `AgentEvent` 转成 NiceEval 的标准事件。
+
+Adapter 只需填写应用地址和审批接口地址。示例中的 `calculate` 工具需要人工审批。
+
+运行评估前，请先用 `pnpm start` 启动应用。NiceEval 不会启动或停止应用进程。
+除依赖声明外，应用的 `src/backend/*` 代码没有改动。
 
 接入的全部代码变更（生成时从两个目录实测统计）：
 
-<table className="gd-summary"><tbody><tr><th>{"类别"}</th><th>{"文件数"}</th><th>{"行数"}</th></tr><tr><td>{"应用侧配置（必要：依赖声明）"}</td><td>{"3"}</td><td>{"+7 −1"}</td></tr><tr><td>{"adapter（必要：传输粘合，协议映射在官方包里）"}</td><td>{"2"}</td><td>{"+110"}</td></tr><tr><td>{"evals 与 experiments（评测内容，按需增长）"}</td><td>{"6"}</td><td>{"+119"}</td></tr><tr className="gd-total"><td>{"合计"}</td><td>{"11"}</td><td>{"+236 −1"}</td></tr></tbody></table>
+<table className="gd-summary"><tbody><tr><th>{"类别"}</th><th>{"文件数"}</th><th>{"行数"}</th></tr><tr><td>{"应用侧配置（必要：依赖声明）"}</td><td>{"3"}</td><td>{"+7 −1"}</td></tr><tr><td>{"Adapter（必要：连接应用，协议映射由 NiceEval 提供）"}</td><td>{"2"}</td><td>{"+110"}</td></tr><tr><td>{"评估用例与实验（按需增长）"}</td><td>{"6"}</td><td>{"+119"}</td></tr><tr className="gd-total"><td>{"合计"}</td><td>{"11"}</td><td>{"+236 −1"}</td></tr></tbody></table>
 
 ## 文件清单
 
@@ -63,7 +65,7 @@ pi-sdk/
 </div>
 </div>
 
-## 新增的 adapter、evals 与 experiments
+## 新增的 Adapter、评估用例与实验
 
 <div className="gd-file">
 <div className="gd-head"><span className="gd-name">{"niceeval.config.ts"}</span><span className="gd-stats"><span className="gd-plus">{"+9"}</span></span></div>
@@ -75,14 +77,14 @@ pi-sdk/
 <div className="gd-file">
 <div className="gd-head"><span className="gd-name">{"agents/pi-sdk.ts"}</span><span className="gd-stats"><span className="gd-plus">{"+101"}</span></span></div>
 <div className="gd-body">
-<table className="gd-table"><tbody><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"1"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// pi-sdk 的 adapter:无侵入对接一个**已经在跑**的应用(../src/backend/server.ts,pi 的原生"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"2"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// `AgentEvent` 原样透传成 SSE,外加三种自定义传输帧:session / approval_request /"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"3"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// server_error,见 server.ts 头注释)。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"4"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"//"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"5"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// `AgentEvent` → 标准事件的映射是官方转换器 `createPiAgentEventStream`(`\"niceeval/adapter\"` 导出)"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"6"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 的事;逐帧驱动 + HITL 挂起用的是官方驱动件 `driveFrameStream`,停轮现场与会话续接的状态槽"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"7"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 都在 `ctx.session` 上——不需要模块级状态；`defineDirectAgent` 只如实声明转换器的证据覆盖。这里只剩传输"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"8"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 粘合:端点在哪、三种传输帧怎么处理、审批打哪个端点——不再手写循环和模块级 Map。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"9"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 无 OTel(pi-agent-core 没有官方集成),事件全部来自转换器。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"10"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"//"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"11"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// HITL:calculate 工具经服务端 beforeToolCall 挂审批。approval_request 帧到达时,流并不关闭——"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"12"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 服务端把执行卡在一个 Promise 上等 POST /api/chat/approve。所以 `driveFrameStream` 在这一帧"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"13"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 返回 `{ pause }`时用 heldSlot 记住\"读了一半的 SSE 流\"(连同转换器状态);下一次 send"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"14"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// (即 t.respond)先打 approve 端点、再继续读同一条流到结束——不重新发 /api/chat。`take(heldSlot)`"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"15"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 取到即清除,一次消费。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"16"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" { completeEvidenceCoverage, createSessionSlot, defineDirectAgent, sseJsonFrames, createPiAgentEventStream, driveFrameStream } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval/adapter\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"17"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"type"}</span><span className="gdt0">{" { AgentContext, PiAgentStream, SseFrameCursor } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval/adapter\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"18"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"type"}</span><span className="gdt0">{" { JsonValue, Turn, TurnInput } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"19"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"type"}</span><span className="gdt0">{" { AgentEvent } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"@earendil-works/pi-agent-core\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"20"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"21"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 被测应用由你自己按它的方式启动(pnpm start / 部署在哪都行),eval 不代管进程、不另开端口。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"22"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"BASE_URL"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" process.env."}</span><span className="gdt1">{"PI_SDK_URL"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"??"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"http://127.0.0.1:33001\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"23"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"24"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"type"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"TransportFrame"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"25"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"|"}</span><span className="gdt0">{" { "}</span><span className="gdt7">{"type"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"session\""}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"sessionId"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{" }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"26"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"|"}</span><span className="gdt0">{" { "}</span><span className="gdt7">{"type"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"approval_request\""}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"toolCallId"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"toolName"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"args"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"unknown"}</span><span className="gdt0">{" }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"27"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"|"}</span><span className="gdt0">{" { "}</span><span className="gdt7">{"type"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"server_error\""}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"message"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{" };"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"28"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"29"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"type"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"PiFrame"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt5">{"AgentEvent"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"|"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"TransportFrame"}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"30"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"31"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"async"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"function"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"appFetch"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"path"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{", "}</span><span className="gdt7">{"body"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"unknown"}</span><span className="gdt0">{", "}</span><span className="gdt7">{"signal"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"AbortSignal"}</span><span className="gdt0">{")"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Promise"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"Response"}</span><span className="gdt0">{"> {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"32"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"try"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"33"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"fetch"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"`${"}</span><span className="gdt1">{"BASE_URL"}</span><span className="gdt2">{"}${"}</span><span className="gdt0">{"path"}</span><span className="gdt2">{"}`"}</span><span className="gdt0">{", {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"34"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      method: "}</span><span className="gdt2">{"\"POST\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"35"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      headers: { "}</span><span className="gdt2">{"\"content-type\""}</span><span className="gdt0">{": "}</span><span className="gdt2">{"\"application/json\""}</span><span className="gdt0">{" },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"36"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      body: "}</span><span className="gdt1">{"JSON"}</span><span className="gdt0">{"."}</span><span className="gdt5">{"stringify"}</span><span className="gdt0">{"(body),"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"37"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      signal,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"38"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"39"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  } "}</span><span className="gdt4">{"catch"}</span><span className="gdt0">{" (err) {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"40"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (signal.aborted) "}</span><span className="gdt4">{"throw"}</span><span className="gdt0">{" err;"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"41"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"throw"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"new"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Error"}</span><span className="gdt0">{"("}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"42"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt2">{"`连不上 ${"}</span><span className="gdt1">{"BASE_URL"}</span><span className="gdt2">{"}${"}</span><span className="gdt0">{"path"}</span><span className="gdt2">{"}。被测应用在跑吗?先起它:cd examples/zh/tier1/pi-sdk && pnpm start(或设 PI_SDK_URL 指向已部署实例)。`"}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"43"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    );"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"44"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"45"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"}"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"46"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"47"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"interface"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Pending"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"48"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"readonly"}</span><span className="gdt0">{" "}</span><span className="gdt7">{"cursor"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"SseFrameCursor"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"PiFrame"}</span><span className="gdt0">{">;"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"49"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"readonly"}</span><span className="gdt0">{" "}</span><span className="gdt7">{"stream"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"PiAgentStream"}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"50"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"readonly"}</span><span className="gdt0">{" "}</span><span className="gdt7">{"toolCallId"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"51"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"}"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"52"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"heldSlot"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt5">{"createSessionSlot"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"Pending"}</span><span className="gdt0">{">("}</span><span className="gdt2">{"\"pi-sdk/held-stream\""}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"53"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"54"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"function"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"readStream"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"cursor"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"SseFrameCursor"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"PiFrame"}</span><span className="gdt0">{">, "}</span><span className="gdt7">{"ctx"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"AgentContext"}</span><span className="gdt0">{", "}</span><span className="gdt7">{"stream"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"PiAgentStream"}</span><span className="gdt0">{")"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Promise"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"Turn"}</span><span className="gdt0">{"> {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"55"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"driveFrameStream"}</span><span className="gdt0">{"(cursor, stream, ctx, ("}</span><span className="gdt7">{"frame"}</span><span className="gdt0">{") "}</span><span className="gdt4">{"=>"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"56"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (frame.type "}</span><span className="gdt4">{"==="}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"session\""}</span><span className="gdt0">{") {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"57"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt6">{"// 会话续接走「服务端记历史」范式:回传的 sessionId 用 ctx.session.capture 写回,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"58"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt6">{"// 只在还没记过时落地(first-writer-wins)。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"59"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      ctx.session."}</span><span className="gdt5">{"capture"}</span><span className="gdt0">{"(frame.sessionId);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"60"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"61"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"62"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (frame.type "}</span><span className="gdt4">{"==="}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"approval_request\""}</span><span className="gdt0">{") {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"63"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      ctx.session."}</span><span className="gdt5">{"set"}</span><span className="gdt0">{"(heldSlot, { cursor, stream, toolCallId: frame.toolCallId });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"64"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"65"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"        pause: { id: frame.toolCallId, action: frame.toolName, input: frame.args "}</span><span className="gdt4">{"as"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"JsonValue"}</span><span className="gdt0">{", options: [{ id: "}</span><span className="gdt2">{"\"approve\""}</span><span className="gdt0">{" }, { id: "}</span><span className="gdt2">{"\"deny\""}</span><span className="gdt0">{" }] },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"66"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      };"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"67"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"68"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (frame.type "}</span><span className="gdt4">{"==="}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"server_error\""}</span><span className="gdt0">{") "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" { fail: frame.message };"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"69"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"70"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"}"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"71"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"72"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"async"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"function"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"send"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"input"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"TurnInput"}</span><span className="gdt0">{", "}</span><span className="gdt7">{"ctx"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"AgentContext"}</span><span className="gdt0">{")"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Promise"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"Turn"}</span><span className="gdt0">{"> {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"73"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"pending"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" ctx.session."}</span><span className="gdt5">{"take"}</span><span className="gdt0">{"(heldSlot);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"74"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (pending) {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"75"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt6">{"// 按 requestId(挂起的 toolCallId)从 input.responses 里对位取裁决,不从 text 猜;"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"76"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt6">{"// 这里每次只挂一条审批,取第一条即可——多请求并停时按 requestId 对位。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"77"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"approved"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" input.responses?.["}</span><span className="gdt1">{"0"}</span><span className="gdt0">{"]?.optionId "}</span><span className="gdt4">{"==="}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"approve\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"78"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" ("}</span><span className="gdt4">{"!"}</span><span className="gdt0">{"approved) pending.stream."}</span><span className="gdt5">{"markRejected"}</span><span className="gdt0">{"(pending.toolCallId);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"79"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"approveRes"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"appFetch"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"/api/chat/approve\""}</span><span className="gdt0">{", { toolUseId: pending.toolCallId, approved }, ctx.signal);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"80"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" ("}</span><span className="gdt4">{"!"}</span><span className="gdt0">{"approveRes.ok) {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"81"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt4">{"throw"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"new"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Error"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"`POST /api/chat/approve 失败: ${"}</span><span className="gdt0">{"approveRes"}</span><span className="gdt2">{"."}</span><span className="gdt0">{"status"}</span><span className="gdt2">{"} ${"}</span><span className="gdt4">{"await"}</span><span className="gdt2">{" "}</span><span className="gdt0">{"approveRes"}</span><span className="gdt2">{"."}</span><span className="gdt5">{"text"}</span><span className="gdt2">{"()"}</span><span className="gdt2">{"}`"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"82"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"83"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"readStream"}</span><span className="gdt0">{"(pending.cursor, ctx, pending.stream);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"84"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"85"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"86"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"res"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"appFetch"}</span><span className="gdt0">{"("}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"87"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt2">{"\"/api/chat\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"88"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    { message: input.text, sessionId: ctx.session.id },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"89"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    ctx.signal,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"90"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  );"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"91"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" ("}</span><span className="gdt4">{"!"}</span><span className="gdt0">{"res.ok "}</span><span className="gdt4">{"||"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"!"}</span><span className="gdt0">{"res.body) {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"92"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"throw"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"new"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Error"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"`POST /api/chat 失败: ${"}</span><span className="gdt0">{"res"}</span><span className="gdt2">{"."}</span><span className="gdt0">{"status"}</span><span className="gdt2">{"} ${"}</span><span className="gdt4">{"await"}</span><span className="gdt2">{" "}</span><span className="gdt0">{"res"}</span><span className="gdt2">{"."}</span><span className="gdt5">{"text"}</span><span className="gdt2">{"()."}</span><span className="gdt5">{"catch"}</span><span className="gdt2">{"(() "}</span><span className="gdt4">{"=>"}</span><span className="gdt2">{" "}</span><span className="gdt2">{"\"\""}</span><span className="gdt2">{")"}</span><span className="gdt2">{"}`"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"93"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"94"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"readStream"}</span><span className="gdt0">{"("}</span><span className="gdt5">{"sseJsonFrames"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"PiFrame"}</span><span className="gdt0">{">(res.body), ctx, "}</span><span className="gdt5">{"createPiAgentEventStream"}</span><span className="gdt0">{"());"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"95"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"}"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"96"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"97"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"export"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"default"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"defineDirectAgent"}</span><span className="gdt0">{"({"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"98"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  name: "}</span><span className="gdt2">{"\"pi-sdk\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"99"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  evidenceCoverage: completeEvidenceCoverage,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"100"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  send,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"101"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"});"}</span></td></tr></tbody></table>
+<table className="gd-table"><tbody><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"1"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// pi-sdk 的 adapter:无侵入对接一个**已经在跑**的应用(../src/backend/server.ts,pi 的原生"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"2"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// `AgentEvent` 原样透传成 SSE,外加三种自定义传输帧:session / approval_request /"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"3"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// server_error,见 server.ts 头注释)。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"4"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"//"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"5"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// `AgentEvent` → 标准事件的映射是官方转换器 `createPiAgentEventStream`(`\"niceeval/adapter\"` 导出)"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"6"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 的事;逐帧驱动 + HITL 挂起用的是官方驱动件 `driveFrameStream`,停轮现场与会话续接的状态槽"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"7"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 都在 `ctx.session` 上——不需要模块级状态；`defineAgent` 只如实声明转换器的证据覆盖。这里只剩传输"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"8"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 粘合:端点在哪、三种传输帧怎么处理、审批打哪个端点——不再手写循环和模块级 Map。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"9"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 无 OTel(pi-agent-core 没有官方集成),事件全部来自转换器。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"10"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"//"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"11"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// HITL:calculate 工具经服务端 beforeToolCall 挂审批。approval_request 帧到达时,流并不关闭——"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"12"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 服务端把执行卡在一个 Promise 上等 POST /api/chat/approve。所以 `driveFrameStream` 在这一帧"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"13"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 返回 `{ pause }`时用 heldSlot 记住\"读了一半的 SSE 流\"(连同转换器状态);下一次 send"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"14"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// (即 t.respond)先打 approve 端点、再继续读同一条流到结束——不重新发 /api/chat。`take(heldSlot)`"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"15"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 取到即清除,一次消费。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"16"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" { completeEvidenceCoverage, createSessionSlot, defineAgent, sseJsonFrames, createPiAgentEventStream, driveFrameStream } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval/adapter\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"17"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"type"}</span><span className="gdt0">{" { AgentContext, PiAgentStream, SseFrameCursor } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval/adapter\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"18"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"type"}</span><span className="gdt0">{" { JsonValue, Turn, TurnInput } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"19"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"type"}</span><span className="gdt0">{" { AgentEvent } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"@earendil-works/pi-agent-core\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"20"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"21"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 被测应用由你自己按它的方式启动(pnpm start / 部署在哪都行),eval 不代管进程、不另开端口。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"22"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"BASE_URL"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" process.env."}</span><span className="gdt1">{"PI_SDK_URL"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"??"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"http://127.0.0.1:33001\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"23"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"24"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"type"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"TransportFrame"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"25"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"|"}</span><span className="gdt0">{" { "}</span><span className="gdt7">{"type"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"session\""}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"sessionId"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{" }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"26"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"|"}</span><span className="gdt0">{" { "}</span><span className="gdt7">{"type"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"approval_request\""}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"toolCallId"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"toolName"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"args"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"unknown"}</span><span className="gdt0">{" }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"27"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"|"}</span><span className="gdt0">{" { "}</span><span className="gdt7">{"type"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"server_error\""}</span><span className="gdt0">{"; "}</span><span className="gdt7">{"message"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{" };"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"28"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"29"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"type"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"PiFrame"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt5">{"AgentEvent"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"|"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"TransportFrame"}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"30"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"31"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"async"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"function"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"appFetch"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"path"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{", "}</span><span className="gdt7">{"body"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"unknown"}</span><span className="gdt0">{", "}</span><span className="gdt7">{"signal"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"AbortSignal"}</span><span className="gdt0">{")"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Promise"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"Response"}</span><span className="gdt0">{"> {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"32"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"try"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"33"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"fetch"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"`${"}</span><span className="gdt1">{"BASE_URL"}</span><span className="gdt2">{"}${"}</span><span className="gdt0">{"path"}</span><span className="gdt2">{"}`"}</span><span className="gdt0">{", {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"34"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      method: "}</span><span className="gdt2">{"\"POST\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"35"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      headers: { "}</span><span className="gdt2">{"\"content-type\""}</span><span className="gdt0">{": "}</span><span className="gdt2">{"\"application/json\""}</span><span className="gdt0">{" },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"36"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      body: "}</span><span className="gdt1">{"JSON"}</span><span className="gdt0">{"."}</span><span className="gdt5">{"stringify"}</span><span className="gdt0">{"(body),"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"37"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      signal,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"38"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"39"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  } "}</span><span className="gdt4">{"catch"}</span><span className="gdt0">{" (err) {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"40"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (signal.aborted) "}</span><span className="gdt4">{"throw"}</span><span className="gdt0">{" err;"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"41"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"throw"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"new"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Error"}</span><span className="gdt0">{"("}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"42"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt2">{"`连不上 ${"}</span><span className="gdt1">{"BASE_URL"}</span><span className="gdt2">{"}${"}</span><span className="gdt0">{"path"}</span><span className="gdt2">{"}。被测应用在跑吗?先起它:cd examples/zh/tier1/pi-sdk && pnpm start(或设 PI_SDK_URL 指向已部署实例)。`"}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"43"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    );"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"44"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"45"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"}"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"46"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"47"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"interface"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Pending"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"48"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"readonly"}</span><span className="gdt0">{" "}</span><span className="gdt7">{"cursor"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"SseFrameCursor"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"PiFrame"}</span><span className="gdt0">{">;"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"49"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"readonly"}</span><span className="gdt0">{" "}</span><span className="gdt7">{"stream"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"PiAgentStream"}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"50"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"readonly"}</span><span className="gdt0">{" "}</span><span className="gdt7">{"toolCallId"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"string"}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"51"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"}"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"52"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"heldSlot"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt5">{"createSessionSlot"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"Pending"}</span><span className="gdt0">{">("}</span><span className="gdt2">{"\"pi-sdk/held-stream\""}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"53"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"54"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"function"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"readStream"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"cursor"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"SseFrameCursor"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"PiFrame"}</span><span className="gdt0">{">, "}</span><span className="gdt7">{"ctx"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"AgentContext"}</span><span className="gdt0">{", "}</span><span className="gdt7">{"stream"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"PiAgentStream"}</span><span className="gdt0">{")"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Promise"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"Turn"}</span><span className="gdt0">{"> {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"55"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"driveFrameStream"}</span><span className="gdt0">{"(cursor, stream, ctx, ("}</span><span className="gdt7">{"frame"}</span><span className="gdt0">{") "}</span><span className="gdt4">{"=>"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"56"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (frame.type "}</span><span className="gdt4">{"==="}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"session\""}</span><span className="gdt0">{") {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"57"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt6">{"// 会话续接走「服务端记历史」范式:回传的 sessionId 用 ctx.session.capture 写回,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"58"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt6">{"// 只在还没记过时落地(first-writer-wins)。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"59"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      ctx.session."}</span><span className="gdt5">{"capture"}</span><span className="gdt0">{"(frame.sessionId);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"60"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"61"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"62"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (frame.type "}</span><span className="gdt4">{"==="}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"approval_request\""}</span><span className="gdt0">{") {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"63"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      ctx.session."}</span><span className="gdt5">{"set"}</span><span className="gdt0">{"(heldSlot, { cursor, stream, toolCallId: frame.toolCallId });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"64"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"65"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"        pause: { id: frame.toolCallId, action: frame.toolName, input: frame.args "}</span><span className="gdt4">{"as"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"JsonValue"}</span><span className="gdt0">{", options: [{ id: "}</span><span className="gdt2">{"\"approve\""}</span><span className="gdt0">{" }, { id: "}</span><span className="gdt2">{"\"deny\""}</span><span className="gdt0">{" }] },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"66"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      };"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"67"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"68"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (frame.type "}</span><span className="gdt4">{"==="}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"server_error\""}</span><span className="gdt0">{") "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" { fail: frame.message };"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"69"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"70"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"}"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"71"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"72"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"async"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"function"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"send"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"input"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"TurnInput"}</span><span className="gdt0">{", "}</span><span className="gdt7">{"ctx"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"AgentContext"}</span><span className="gdt0">{")"}</span><span className="gdt4">{":"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Promise"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"Turn"}</span><span className="gdt0">{"> {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"73"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"pending"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" ctx.session."}</span><span className="gdt5">{"take"}</span><span className="gdt0">{"(heldSlot);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"74"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" (pending) {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"75"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt6">{"// 按 requestId(挂起的 toolCallId)从 input.responses 里对位取裁决,不从 text 猜;"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"76"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt6">{"// 这里每次只挂一条审批,取第一条即可——多请求并停时按 requestId 对位。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"77"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"approved"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" input.responses?.["}</span><span className="gdt1">{"0"}</span><span className="gdt0">{"]?.optionId "}</span><span className="gdt4">{"==="}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"approve\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"78"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" ("}</span><span className="gdt4">{"!"}</span><span className="gdt0">{"approved) pending.stream."}</span><span className="gdt5">{"markRejected"}</span><span className="gdt0">{"(pending.toolCallId);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"79"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"approveRes"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"appFetch"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"/api/chat/approve\""}</span><span className="gdt0">{", { toolUseId: pending.toolCallId, approved }, ctx.signal);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"80"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" ("}</span><span className="gdt4">{"!"}</span><span className="gdt0">{"approveRes.ok) {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"81"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      "}</span><span className="gdt4">{"throw"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"new"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Error"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"`POST /api/chat/approve 失败: ${"}</span><span className="gdt0">{"approveRes"}</span><span className="gdt2">{"."}</span><span className="gdt0">{"status"}</span><span className="gdt2">{"} ${"}</span><span className="gdt4">{"await"}</span><span className="gdt2">{" "}</span><span className="gdt0">{"approveRes"}</span><span className="gdt2">{"."}</span><span className="gdt5">{"text"}</span><span className="gdt2">{"()"}</span><span className="gdt2">{"}`"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"82"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"83"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"readStream"}</span><span className="gdt0">{"(pending.cursor, ctx, pending.stream);"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"84"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"85"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"86"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"res"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"appFetch"}</span><span className="gdt0">{"("}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"87"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt2">{"\"/api/chat\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"88"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    { message: input.text, sessionId: ctx.session.id },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"89"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    ctx.signal,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"90"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  );"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"91"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"if"}</span><span className="gdt0">{" ("}</span><span className="gdt4">{"!"}</span><span className="gdt0">{"res.ok "}</span><span className="gdt4">{"||"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"!"}</span><span className="gdt0">{"res.body) {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"92"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"throw"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"new"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"Error"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"`POST /api/chat 失败: ${"}</span><span className="gdt0">{"res"}</span><span className="gdt2">{"."}</span><span className="gdt0">{"status"}</span><span className="gdt2">{"} ${"}</span><span className="gdt4">{"await"}</span><span className="gdt2">{" "}</span><span className="gdt0">{"res"}</span><span className="gdt2">{"."}</span><span className="gdt5">{"text"}</span><span className="gdt2">{"()."}</span><span className="gdt5">{"catch"}</span><span className="gdt2">{"(() "}</span><span className="gdt4">{"=>"}</span><span className="gdt2">{" "}</span><span className="gdt2">{"\"\""}</span><span className="gdt2">{")"}</span><span className="gdt2">{"}`"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"93"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  }"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"94"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"return"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"readStream"}</span><span className="gdt0">{"("}</span><span className="gdt5">{"sseJsonFrames"}</span><span className="gdt0">{"<"}</span><span className="gdt5">{"PiFrame"}</span><span className="gdt0">{">(res.body), ctx, "}</span><span className="gdt5">{"createPiAgentEventStream"}</span><span className="gdt0">{"());"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"95"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"}"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"96"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"97"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"export"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"default"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"defineAgent"}</span><span className="gdt0">{"({"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"98"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  name: "}</span><span className="gdt2">{"\"pi-sdk\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"99"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  evidenceCoverage: completeEvidenceCoverage,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"100"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  send,"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"101"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"});"}</span></td></tr></tbody></table>
 </div>
 </div>
 
 <div className="gd-file">
 <div className="gd-head"><span className="gd-name">{"evals/basic-qa.eval.ts"}</span><span className="gd-stats"><span className="gd-plus">{"+22"}</span></span></div>
 <div className="gd-body">
-<table className="gd-table"><tbody><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"1"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" { defineEval } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"2"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"3"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 这条 eval 验证 agent 能正常问答、不瞎调工具,顺带冒烟 usage 有没有正确从"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"4"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// message_end 的 AssistantMessage.usage 累加进 Turn.usage。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"5"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"export"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"default"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"defineEval"}</span><span className="gdt0">{"({"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"6"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  description: "}</span><span className="gdt2">{"\"测试 agent 能正常问答且不瞎调工具\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"7"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"8"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"async"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"test"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"t"}</span><span className="gdt0">{") {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"9"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"turn"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" t."}</span><span className="gdt5">{"send"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"用一句话介绍一下你自己,这轮不用查天气也不用算数。\""}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"10"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" turn."}</span><span className="gdt5">{"succeeded"}</span><span className="gdt0">{"()."}</span><span className="gdt5">{"stopOnFailure"}</span><span className="gdt0">{"();"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"11"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"12"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" t."}</span><span className="gdt5">{"group"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"正常收发、没有多余工具调用\""}</span><span className="gdt0">{", () "}</span><span className="gdt4">{"=>"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"13"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      t."}</span><span className="gdt5">{"succeeded"}</span><span className="gdt0">{"();"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"14"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      t."}</span><span className="gdt5">{"usedNoTools"}</span><span className="gdt0">{"();"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"15"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"16"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"17"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt6">{"// usage 冒烟:拿不到就应该是 0,不该抛错;上限给得宽松,只为证明数字不是编的。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"18"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    t."}</span><span className="gdt5">{"maxTokens"}</span><span className="gdt0">{"("}</span><span className="gdt1">{"20_000"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"19"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"20"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    t.judge.autoevals."}</span><span className="gdt5">{"closedQA"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"助手是否用一两句话正常介绍了自己,而不是报错或答非所问?\""}</span><span className="gdt0">{")."}</span><span className="gdt5">{"gate"}</span><span className="gdt0">{"("}</span><span className="gdt1">{"0.6"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"21"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"22"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"});"}</span></td></tr></tbody></table>
+<table className="gd-table"><tbody><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"1"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" { defineEval } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"2"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"3"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 这条 eval 验证 agent 能正常问答、不瞎调工具,顺带冒烟 usage 有没有正确从"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"4"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// message_end 的 AssistantMessage.usage 累加进 Turn.usage。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"5"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"export"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"default"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"defineEval"}</span><span className="gdt0">{"({"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"6"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  description: "}</span><span className="gdt2">{"\"测试 agent 能正常问答且不瞎调工具\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"7"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"8"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"async"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"test"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"t"}</span><span className="gdt0">{") {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"9"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"turn"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" t."}</span><span className="gdt5">{"send"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"用一句话介绍一下你自己,这轮不用查天气也不用算数。\""}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"10"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" turn."}</span><span className="gdt5">{"succeeded"}</span><span className="gdt0">{"()."}</span><span className="gdt5">{"orStop"}</span><span className="gdt0">{"();"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"11"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"12"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" t."}</span><span className="gdt5">{"group"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"正常收发、没有多余工具调用\""}</span><span className="gdt0">{", () "}</span><span className="gdt4">{"=>"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"13"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      t."}</span><span className="gdt5">{"succeeded"}</span><span className="gdt0">{"();"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"14"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      t."}</span><span className="gdt5">{"usedNoTools"}</span><span className="gdt0">{"();"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"15"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"16"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"17"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt6">{"// usage 冒烟:拿不到就应该是 0,不该抛错;上限给得宽松,只为证明数字不是编的。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"18"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    t."}</span><span className="gdt5">{"maxTokens"}</span><span className="gdt0">{"("}</span><span className="gdt1">{"20_000"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"19"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"20"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    t.judge.autoevals."}</span><span className="gdt5">{"closedQA"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"助手是否用一两句话正常介绍了自己,而不是报错或答非所问?\""}</span><span className="gdt0">{")."}</span><span className="gdt5">{"gate"}</span><span className="gdt0">{"("}</span><span className="gdt1">{"0.6"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"21"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"22"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"});"}</span></td></tr></tbody></table>
 </div>
 </div>
 
@@ -110,7 +112,7 @@ pi-sdk/
 <div className="gd-file">
 <div className="gd-head"><span className="gd-name">{"evals/weather-tool.eval.ts"}</span><span className="gd-stats"><span className="gd-plus">{"+21"}</span></span></div>
 <div className="gd-body">
-<table className="gd-table"><tbody><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"1"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" { defineEval } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"2"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"3"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 这条 eval 验证 agent 遇到实时天气问题时会调 get_weather,而不是直接编一个答案。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"4"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"export"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"default"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"defineEval"}</span><span className="gdt0">{"({"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"5"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  description: "}</span><span className="gdt2">{"\"测试 agent 在天气问题中正确调用 get_weather 并基于结果作答\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"6"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"7"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"async"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"test"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"t"}</span><span className="gdt0">{") {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"8"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"turn"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" t."}</span><span className="gdt5">{"send"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"北京今天天气怎么样?\""}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"9"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" turn."}</span><span className="gdt5">{"succeeded"}</span><span className="gdt0">{"()."}</span><span className="gdt5">{"stopOnFailure"}</span><span className="gdt0">{"();"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"10"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"11"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" t."}</span><span className="gdt5">{"group"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"调用 get_weather 且城市正确\""}</span><span className="gdt0">{", () "}</span><span className="gdt4">{"=>"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"12"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      t."}</span><span className="gdt5">{"calledTool"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"get_weather\""}</span><span className="gdt0">{", { input: { city: "}</span><span className="gdt2">{"\"北京\""}</span><span className="gdt0">{" } });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"13"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      t."}</span><span className="gdt5">{"messageIncludes"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"/"}</span><span className="gdt8">{"°C"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"气温"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"天气"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"晴"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"多云"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"雨"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"阴"}</span><span className="gdt2">{"/"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"14"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"15"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"16"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt6">{"// 「是否走了工具」由上面的 t.calledTool 确定性把关;judge 只评回复本身的质量。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"17"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    t.judge.autoevals"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"18"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      ."}</span><span className="gdt5">{"closedQA"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"助手是否给出了具体的天气数据(温度或天气状况),而不是拒绝回答或含糊其辞?\""}</span><span className="gdt0">{")"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"19"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      ."}</span><span className="gdt5">{"atLeast"}</span><span className="gdt0">{"("}</span><span className="gdt1">{"0.7"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"20"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"21"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"});"}</span></td></tr></tbody></table>
+<table className="gd-table"><tbody><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"1"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"import"}</span><span className="gdt0">{" { defineEval } "}</span><span className="gdt4">{"from"}</span><span className="gdt0">{" "}</span><span className="gdt2">{"\"niceeval\""}</span><span className="gdt0">{";"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"2"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"3"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt6">{"// 这条 eval 验证 agent 遇到实时天气问题时会调 get_weather,而不是直接编一个答案。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"4"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt4">{"export"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"default"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"defineEval"}</span><span className="gdt0">{"({"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"5"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  description: "}</span><span className="gdt2">{"\"测试 agent 在天气问题中正确调用 get_weather 并基于结果作答\""}</span><span className="gdt0">{","}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"6"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"7"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  "}</span><span className="gdt4">{"async"}</span><span className="gdt0">{" "}</span><span className="gdt5">{"test"}</span><span className="gdt0">{"("}</span><span className="gdt7">{"t"}</span><span className="gdt0">{") {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"8"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"const"}</span><span className="gdt0">{" "}</span><span className="gdt1">{"turn"}</span><span className="gdt0">{" "}</span><span className="gdt4">{"="}</span><span className="gdt0">{" "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" t."}</span><span className="gdt5">{"send"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"北京今天天气怎么样?\""}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"9"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" turn."}</span><span className="gdt5">{"succeeded"}</span><span className="gdt0">{"()."}</span><span className="gdt5">{"orStop"}</span><span className="gdt0">{"();"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"10"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"11"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt4">{"await"}</span><span className="gdt0">{" t."}</span><span className="gdt5">{"group"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"调用 get_weather 且城市正确\""}</span><span className="gdt0">{", () "}</span><span className="gdt4">{"=>"}</span><span className="gdt0">{" {"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"12"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      t."}</span><span className="gdt5">{"calledTool"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"get_weather\""}</span><span className="gdt0">{", { input: { city: "}</span><span className="gdt2">{"\"北京\""}</span><span className="gdt0">{" } });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"13"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      t."}</span><span className="gdt5">{"messageIncludes"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"/"}</span><span className="gdt8">{"°C"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"气温"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"天气"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"晴"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"多云"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"雨"}</span><span className="gdt4">{"|"}</span><span className="gdt8">{"阴"}</span><span className="gdt2">{"/"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"14"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    });"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"15"}</td><td className="gd-sign">{"+"}</td><td className="gd-code">{" "}</td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"16"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    "}</span><span className="gdt6">{"// 「是否走了工具」由上面的 t.calledTool 确定性把关;judge 只评回复本身的质量。"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"17"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"    t.judge.autoevals"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"18"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      ."}</span><span className="gdt5">{"closedQA"}</span><span className="gdt0">{"("}</span><span className="gdt2">{"\"助手是否给出了具体的天气数据(温度或天气状况),而不是拒绝回答或含糊其辞?\""}</span><span className="gdt0">{")"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"19"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"      ."}</span><span className="gdt5">{"atLeast"}</span><span className="gdt0">{"("}</span><span className="gdt1">{"0.7"}</span><span className="gdt0">{");"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"20"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"  },"}</span></td></tr><tr className="gd-add"><td className="gd-ln"></td><td className="gd-ln">{"21"}</td><td className="gd-sign">{"+"}</td><td className="gd-code"><span className="gdt0">{"});"}</span></td></tr></tbody></table>
 </div>
 </div>