From 5b5afbc92b8d24bb75dd87dada742da7bd750505 Mon Sep 17 00:00:00 2001 From: Lyon <88232613+pikasTech@users.noreply.github.com> Date: Tue, 23 Jun 2026 08:07:21 +0800 Subject: [PATCH] fix(web-probe): accept stdin prompts for observe commands (#724) Co-authored-by: Codex --- .agents/skills/unidesk-webdev/SKILL.md | 5 ++++- scripts/src/hwlab-node-help.ts | 3 ++- scripts/src/hwlab-node-impl.ts | 13 +++++++++++-- 3 files changed, 17 insertions(+), 4 deletions(-) diff --git a/.agents/skills/unidesk-webdev/SKILL.md b/.agents/skills/unidesk-webdev/SKILL.md index f3ee336a..5e5b2805 100644 --- a/.agents/skills/unidesk-webdev/SKILL.md +++ b/.agents/skills/unidesk-webdev/SKILL.md @@ -84,6 +84,9 @@ bun scripts/cli.ts hwlab nodes web-probe observe start --node D601 --lane v03 -- bun scripts/cli.ts hwlab nodes web-probe observe command webobs-xxxx --type newSession bun scripts/cli.ts hwlab nodes web-probe observe command webobs-xxxx --type selectProvider --provider codex-api bun scripts/cli.ts hwlab nodes web-probe observe command webobs-xxxx --type sendPrompt --text 'ping' +bun scripts/cli.ts hwlab nodes web-probe observe command webobs-xxxx --type sendPrompt --text-stdin <<'EOF' +long prompt +EOF bun scripts/cli.ts hwlab nodes web-probe observe status webobs-xxxx --tail-lines 6 bun scripts/cli.ts hwlab nodes web-probe observe stop webobs-xxxx bun scripts/cli.ts hwlab nodes web-probe observe analyze webobs-xxxx @@ -94,7 +97,7 @@ bun scripts/cli.ts hwlab nodes web-probe observe analyze webobs-xxxx 约束： - `web-probe script` 不运行默认探针，必须通过 stdin heredoc 或 `--script-file ` 提供脚本；只需要 repo-owned 标准 DOM probe 时使用 `web-probe run`。 -- `web-probe observe start` 默认是被动观测：记录 DOM 摘要、自然页面 request/response/requestfailed、截图和 performance 样本，不主动 fetch Workbench API、不切换 control session、不拦截路由、不调用 repair helper。长程 Workbench 观测必须保留 control/observer 双页面模型：control 页面执行显式 command，observer 页面只同步到同一 session URL 后被动采样，并按默认 180000ms 周期整页刷新同一 session 来模拟用户往返；周期刷新只作用于 observer，不得改变 control active session 或作为通过条件。两页的 `pageRole`、`pageId`、`sampleGroupSeq` 必须进入样本和 analyzer 报表。任何 `newSession`、`selectProvider`、`sendPrompt`、`goto`、`screenshot`、`mark`、`stop` 都必须通过 `observe command` 显式下发，并进入 `control.jsonl`。 +- `web-probe observe start` 默认是被动观测：记录 DOM 摘要、自然页面 request/response/requestfailed、截图和 performance 样本，不主动 fetch Workbench API、不切换 control session、不拦截路由、不调用 repair helper。长程 Workbench 观测必须保留 control/observer 双页面模型：control 页面执行显式 command，observer 页面只同步到同一 session URL 后被动采样，并按默认 180000ms 周期整页刷新同一 session 来模拟用户往返；周期刷新只作用于 observer，不得改变 control active session 或作为通过条件。两页的 `pageRole`、`pageId`、`sampleGroupSeq` 必须进入样本和 analyzer 报表。任何 `newSession`、`selectProvider`、`sendPrompt`、`goto`、`screenshot`、`mark`、`stop` 都必须通过 `observe command` 显式下发，并进入 `control.jsonl`；长 prompt 必须优先用 `sendPrompt --text-stdin`，不要为了绕开 shell quoting 退回裸 Playwright 或临时脚本。 - `web-probe observe` 的 issue evidence 优先记录 observer id、stateDir、report JSON/Markdown SHA、samples/control/network/artifact 计数、routeSessionId、activeSessionId、prompt hash/textBytes、traceId、AgentRun runId/commandId、最终 status 和必要摘要；不要把 prompt 原文、assistant 大段正文、完整 stdout/stderr 或 provider payload 粘贴到 issue。 - 多轮 Workbench 采样必须证明同一个 `sessionId` 连续承载所有轮次；每轮至少记录 prompt hash、traceId、终态、最终回答摘要和性能/产物表。若 Web UI 投影卡住但 Code Agent/AgentRun result 已 terminal，应同时登记“执行终态”和“Workbench 投影未收敛”，不得用 `goto`、reload、切 session 或 result polling 把 UI 失败伪装成通过。 - `observe analyze` 是离线分析，只读取 artifact JSONL 并写 `analysis/report.md` 与 `analysis/report.json`，不访问 Workbench API、不驱动浏览器。`observe start` 每次启动必须先把同一 stateDir 中已有的根目录 JSONL 轮转到带时间戳的 `archive/` 文件；`observe analyze` 默认只分析当前根目录 JSONL，不扫描历史 archive，只有显式指定 archive prefix 时才分析历史轮转窗口。报告必须输出采样点 vs 每个 turn 的总耗时/最近更新时间表、可见“加载中”的数量/归属/并发 owner/连续出现区间、DOM diagnostic/HTTP/console/requestfailed/runtime execution error 分组、page asset provenance segment、同源 API Resource Timing 分位表和超过 5s 的慢路径 finding；页面/API 加载超过 5s 视为不可用级性能红线，可见“加载中”持续超过 5s 也必须作为真实慢加载证据登记到上游问题。修复必须降低真实请求、投影、渲染或后端路径耗时，禁止为了减少“加载中”出现时间而提前展示未加载完的内容，也不能靠下游 retry/reload/fallback 掩盖。报告里的 `final-response-flicker`、`uncommanded-visible-state-change`、session changed、network 503 等 finding 是排障线索；用于 closeout 时必须结合原始 session/trace/DOM 证据解释，避免把采样噪声直接当作业务结论。 diff --git a/scripts/src/hwlab-node-help.ts b/scripts/src/hwlab-node-help.ts index 67bdc1cb..6714bdf1 100644 --- a/scripts/src/hwlab-node-help.ts +++ b/scripts/src/hwlab-node-help.ts @@ -79,6 +79,7 @@ export function hwlabNodeWebProbeHelp(): Record { "bun scripts/cli.ts hwlab nodes web-probe observe command webobs-xxxx --type newSession", "bun scripts/cli.ts hwlab nodes web-probe observe command webobs-xxxx --type selectProvider --provider codex-api", "bun scripts/cli.ts hwlab nodes web-probe observe command webobs-xxxx --type sendPrompt --text 'ping'", + "bun scripts/cli.ts hwlab nodes web-probe observe command webobs-xxxx --type sendPrompt --text-stdin <<'EOF'\nlong prompt\nEOF", "bun scripts/cli.ts hwlab nodes web-probe observe status webobs-xxxx", "bun scripts/cli.ts hwlab nodes web-probe observe analyze webobs-xxxx", "bun scripts/cli.ts hwlab nodes web-probe script --node D601 --lane v03 <<'JS'\nexport default async ({ waitWorkbenchReady, fetchJson, fetchApiMatrix, recordStep, collectText, safeEvaluate, screenshot }) => {\n const ready = await waitWorkbenchReady();\n const workspace = await fetchJson('/v1/workbench/workspace?projectId=prj_hwpod_workbench');\n const apiMatrix = await fetchApiMatrix(['/v1/workbench/workspace?projectId=prj_hwpod_workbench', '/auth/session']);\n const workspaceText = await collectText('#workspace');\n const evaluated = await safeEvaluate(({ a, b }) => ({ sum: a + b }), { a: 1, b: 2 });\n await screenshot('workbench.png');\n recordStep('workbench-summary', { finalUrl: ready.finalUrl, workspaceOk: workspace.ok, apiMatrixOk: apiMatrix.ok });\n return { finalUrl: ready.finalUrl, workspaceOk: workspace.ok, workspaceText, evaluated };\n};\nJS", @@ -93,7 +94,7 @@ export function hwlabNodeWebProbeHelp(): Record { "Issue-ready evidence is available under issueEvidence and summary.issueEvidence; full script report is persisted under probe.reportPath with a SHA-256 fingerprint.", "observe sampling is passive by default: it records DOM summaries and natural page request/response/requestfailed events with observerInitiated=false; it does not actively fetch Workbench APIs, reload, switch sessions, route/intercept, or call repair helpers.", "observe start registers a local UniDesk-side observer id under .state/web-observe/index.json; after start, prefer observe status|command|stop|collect|analyze instead of repeating --node/--lane/--state-dir.", - "observe command actions are explicit user/control actions and are appended to control.jsonl; use --type newSession/selectProvider/sendPrompt/goto/screenshot/mark/stop and keep prompt text out of issue comments by citing textHash/textBytes.", + "observe command actions are explicit user/control actions and are appended to control.jsonl; use --type newSession/selectProvider/sendPrompt/goto/screenshot/mark/stop. For long prompts use sendPrompt --text-stdin; keep prompt text out of issue comments by citing textHash/textBytes.", "observe analyze is offline-only: it reads artifact JSONL and writes analysis/report.md plus analysis/report.json without accessing Workbench APIs or driving the browser.", "observe analyze scans every sampled DOM point, extracts Workbench timing text such as 总耗时/total and 最近 N 秒/分前, and writes a sample point vs turn timing report: each Markdown table row starts with the timestamp, followed by each turn's 总耗时(s) and 最近更新(s). Timing series are reported for post-processing/manual analysis instead of auto-judged from status tail output.", "observe analyze also reports visible “加载中” count, owner attribution, concurrent loading owners, and continuous visible segments; fixes must reduce real loading latency, not reveal incomplete content early to make this metric disappear.", diff --git a/scripts/src/hwlab-node-impl.ts b/scripts/src/hwlab-node-impl.ts index 5d21df45..db4a5f77 100644 --- a/scripts/src/hwlab-node-impl.ts +++ b/scripts/src/hwlab-node-impl.ts @@ -6923,7 +6923,7 @@ function parseNodeWebProbeObserveOptions( "--label", "--session-id", "--provider", - ]), new Set(["--force", "--full"])); + ]), new Set(["--force", "--full", "--text-stdin"])); const commandTypeRaw = optionValue(args, "--type") ?? null; const commandType = commandTypeRaw === null ? null : parseNodeWebProbeObserveCommandType(commandTypeRaw); const stateDir = optionValue(args, "--state-dir") ?? indexed?.stateDir ?? null; @@ -6937,6 +6937,15 @@ function parseNodeWebProbeObserveOptions( if (observeActionRaw !== "start" && stateDir === null && jobId === null) { throw new Error("web-probe observe status|command|stop|collect|analyze requires --state-dir or --job-id"); } + const commandTextOption = optionValue(args, "--text") ?? null; + const commandTextFromStdin = args.includes("--text-stdin"); + if (commandTextFromStdin && observeActionRaw !== "command") { + throw new Error("web-probe observe --text-stdin is only supported for observe command"); + } + if (commandTextFromStdin && commandTextOption !== null) { + throw new Error("web-probe observe command accepts either --text or --text-stdin, not both"); + } + const commandText = commandTextFromStdin ? readFileSync(0, "utf8") : commandTextOption; return { action: "observe", observeAction: observeActionRaw, @@ -6962,7 +6971,7 @@ function parseNodeWebProbeObserveOptions( jobId, force: args.includes("--force"), commandType, - commandText: optionValue(args, "--text") ?? null, + commandText, commandPath: optionValue(args, "--path") ?? null, commandLabel: optionValue(args, "--label") ?? null, commandSessionId: optionValue(args, "--session-id") ?? null,