@tea-agent/loop-agent 0.24.7 → 0.24.9

package/CHANGELOG.md

@@ -2,13 +2,36 @@
 
 ## [Unreleased]
 
+## [0.24.9] - 2026-07-29
+
+### 重点更新
+
+- 后端测试执行失败时的诊断信息更加完善，避免出现无输出导致无法定位的问题
+- 修复 Windows 环境下前端证据校验因长命令终端转义导致的崩溃
+- 节点检查器新增「节点输入」只读页签，支持安全审查冻结的节点定义
+
 ### 新增
 
-- Observe / Inspect 节点检查器在「节点输出」左侧新增「节点输入」页签：只读投影冻结 `run.json` 顶层节点定义（任务正文、依赖、执行器摘要、边界）与可选 assembled prompt 指纹；完整 assembled prompt 与动态展开子节点仍不在本轮范围
+- Observe / Inspect 节点检查器在「节点输出」左侧新增「节点输入」页签：只读投影冻结 run.json 顶层节点定义（任务正文、依赖、执行器摘要、边界）与可选 assembled prompt 指纹
+
+### 改进
+
+- 后端测试节点执行失败时可观测性增强：写入包含映射脚本、PYTHON_BIN、pytest 版本、exit、html 是否存在与 STEP 等信息的诊断文件，并在节点 stderr 汇总失败摘要
+
+### 修复
+
+- 修复前端证据校验在 Windows 下通过长 node -e 命令执行时的终端转义崩溃：改为由 runtime 内部直接校验，兼容 CMD、Git Bash、PowerShell 与 Unix shell
+
+## [0.24.8] - 2026-07-29
+
+### 重点更新
+
+- 修复 FINAL-VERIFY 改写已存在的 allowed_paths 证据（gitignore）时 Git checkpoint 误失败，以及 record-error 后 Task Pool 卡在 Running 无法 mark-failed/retry 的问题
 
 ### 修复
 
-- 修复前端证据校验在 Windows 下通过长 `node -e` 命令执行时的终端转义崩溃：改为由 runtime 内部直接校验，兼容 CMD、Git Bash、PowerShell 与 Unix shell
+- `finalizeGitTask` 对 allowed_paths 内的 pre-existing ignored evidence 允许 rewrite 并 force-add
+- `runReadyTasks` 在 record-error 时 best-effort 投影 `runs.jsonl` + Failed 状态，便于官方 retry
 
 ## [0.24.7] - 2026-07-29
 

package/dist/executors/shell-executor.js

@@ -385,28 +385,101 @@ async function executeBackendTestPipeline(input, meta) {
             const mappedScripts = await collectBackendTestMappedPytestScripts(input.cwd);
             const shellQuote = (value) => `'${value.replaceAll("'", `'"'"'`)}'`;
             const pytestTargets = mappedScripts.map(shellQuote).join(" ");
+            const mappedScriptDiagLines = mappedScripts
+                .map((script) => `echo "  - ${script.replaceAll('"', '\\"')}" >> "\${DIAG_FILE}"`)
+                .join("; ");
+            // Diagnostic-first single shell: keep pytest exactly once, but always
+            // emit STEP markers, mapped targets, interpreter identity, exit code
+            // and html presence so empty nonzero-exit failures remain auditable.
             const pytestCommand = [
                 'mkdir -p "${HARNESS_DAG_RUN_DIR}/reports"',
+                'DIAG_FILE="${HARNESS_DAG_RUN_DIR}/reports/backend-test-execute-diagnostics.md"',
+                [
+                    'echo "# Backend Test Execute Diagnostics" > "${DIAG_FILE}"',
+                    'echo >> "${DIAG_FILE}"',
+                    'echo "- startedAt: $(date -u +%Y-%m-%dT%H:%M:%SZ 2>/dev/null || date)" >> "${DIAG_FILE}"',
+                    'echo "- pwd: $(pwd)" >> "${DIAG_FILE}"',
+                    'echo "- HARNESS_DAG_RUN_DIR: ${HARNESS_DAG_RUN_DIR}" >> "${DIAG_FILE}"',
+                    'echo "- shell: ${BASH:-bash} (${BASH_VERSION:-unknown})" >> "${DIAG_FILE}"',
+                    `echo "- mappedScriptCount: ${mappedScripts.length}" >> "\${DIAG_FILE}"`,
+                    'echo "- mappedScripts:" >> "${DIAG_FILE}"',
+                ].join("; "),
+                mappedScriptDiagLines || 'echo "  - <none>" >> "${DIAG_FILE}"',
+                'echo "STEP=resolve-python"',
                 'PYTHON_BIN="$(command -v python || command -v python3 || true)"',
-                'if [ -z "${PYTHON_BIN}" ]; then echo "python/python3 is required for backend-test execution" >&2; exit 127; fi',
+                [
+                    'echo "- command_v_python: $(command -v python || true)" >> "${DIAG_FILE}"',
+                    'echo "- command_v_python3: $(command -v python3 || true)" >> "${DIAG_FILE}"',
+                    'echo "- PYTHON_BIN: ${PYTHON_BIN:-<empty>}" >> "${DIAG_FILE}"',
+                ].join("; "),
+                'if [ -z "${PYTHON_BIN}" ]; then echo "python/python3 is required for backend-test execution" | tee -a "${DIAG_FILE}" >&2; exit 127; fi',
+                'echo "STEP=python-identity"',
+                'if ! PYTHONUNBUFFERED=1 "${PYTHON_BIN}" -c "import sys; print(sys.executable); print(sys.version.splitlines()[0])" >> "${DIAG_FILE}" 2>&1; then echo "STEP=python-identity-failed" | tee -a "${DIAG_FILE}" >&2; exit 127; fi',
+                'echo "STEP=pytest-version"',
+                'if ! PYTHONUNBUFFERED=1 "${PYTHON_BIN}" -m pytest --version >> "${DIAG_FILE}" 2>&1; then echo "STEP=pytest-version-failed" | tee -a "${DIAG_FILE}" >&2; exit 127; fi',
+                'echo "STEP=pytest-run"',
+                `echo "- pytestCommand: python -m pytest ${mappedScripts.join(" ")} -v -p no:cacheprovider --html=reports/backend-test.html --self-contained-html" >> "\${DIAG_FILE}"`,
                 // Run pytest exactly once and emit the native pytest-html
                 // self-contained report. The HTML/facts renderers read per-case
                 // captured stdout (HTTP_REQUEST/HTTP_RESPONSE) directly from the
                 // pytest-html data-jsonblob island, so a JUnit XML report is no
                 // longer generated.
-                `PYTHONUTF8=1 PYTHONIOENCODING=utf-8 PYTHONDONTWRITEBYTECODE=1 "\${PYTHON_BIN}" -m pytest ${pytestTargets} -v -p no:cacheprovider --html="\${HARNESS_DAG_RUN_DIR}/reports/backend-test.html" --self-contained-html`,
+                `PYTHONUTF8=1 PYTHONIOENCODING=utf-8 PYTHONDONTWRITEBYTECODE=1 PYTHONUNBUFFERED=1 "\${PYTHON_BIN}" -m pytest ${pytestTargets} -v -p no:cacheprovider --html="\${HARNESS_DAG_RUN_DIR}/reports/backend-test.html" --self-contained-html`,
                 "STATUS=$?",
                 'printf "%s" "${STATUS}" > "${HARNESS_DAG_RUN_DIR}/reports/backend-test-pytest-exit.txt"',
+                [
+                    'echo "- pytestExitCode: ${STATUS}" >> "${DIAG_FILE}"',
+                    'if [ -f "${HARNESS_DAG_RUN_DIR}/reports/backend-test.html" ]; then echo "- pytestHtml: present ($(wc -c < "${HARNESS_DAG_RUN_DIR}/reports/backend-test.html" | tr -d " ") bytes)" >> "${DIAG_FILE}"; else echo "- pytestHtml: missing" >> "${DIAG_FILE}"; fi',
+                    'echo "- finishedAt: $(date -u +%Y-%m-%dT%H:%M:%SZ 2>/dev/null || date)" >> "${DIAG_FILE}"',
+                ].join("; "),
+                'echo "STEP=pytest-finished status=${STATUS}"',
                 // exit 0 (all pass) or 1 (assertion failures) with a valid
                 // pytest-html report are reportable; collection errors / crashes
                 // (exit >= 2) or a missing report surface as a pipeline error below.
                 'if { [ "${STATUS}" -eq 0 ] || [ "${STATUS}" -eq 1 ]; } && [ -s "${HARNESS_DAG_RUN_DIR}/reports/backend-test.html" ]; then exit 0; fi',
+                'echo "STEP=pipeline-failed status=${STATUS} html=$([ -s "${HARNESS_DAG_RUN_DIR}/reports/backend-test.html" ] && echo present || echo missing)" >&2',
                 'exit "${STATUS}"',
             ].join("; ");
             const results = await executePipelineCommands(input, meta, [pytestCommand]);
             if (!results.every((result) => result.ok)) {
                 const failure = results.find((result) => !result.ok);
-                return { ok: false, stdout: results.map((result) => result.stdout).join("\n"), stderr: failure.stderr, failureCategory: failure.failureCategory, durationMs: Date.now() - started };
+                const combinedStdout = results.map((result) => result.stdout).join("\n");
+                let diagnosticsText = "";
+                try {
+                    diagnosticsText = await readFile(path.join(meta.runDir, "reports", "backend-test-execute-diagnostics.md"), "utf8");
+                }
+                catch {
+                    diagnosticsText = "";
+                }
+                const stderrSummary = [
+                    "backend-test markdown-execute-html shell failed",
+                    `failureCategory=${failure.failureCategory}`,
+                    `exitCode=${failure.exitCode ?? "null"}`,
+                    `durationMs=${failure.durationMs}`,
+                    `mappedScripts=${mappedScripts.join(",") || "<none>"}`,
+                    `shellStdoutEmpty=${combinedStdout.trim().length === 0}`,
+                    `shellStderrEmpty=${failure.stderr.trim().length === 0}`,
+                    failure.stdoutArtifactPath
+                        ? `commandStdoutArtifact=${failure.stdoutArtifactPath}`
+                        : undefined,
+                    failure.stderrArtifactPath
+                        ? `commandStderrArtifact=${failure.stderrArtifactPath}`
+                        : undefined,
+                    "diagnostics=reports/backend-test-execute-diagnostics.md",
+                    failure.stderr.trim() || "(shell stderr empty)",
+                    diagnosticsText.trim()
+                        ? `--- diagnostics ---\n${diagnosticsText.trim()}`
+                        : "(diagnostics file missing or empty)",
+                ]
+                    .filter((line) => Boolean(line))
+                    .join("\n");
+                return {
+                    ok: false,
+                    stdout: combinedStdout,
+                    stderr: stderrSummary,
+                    failureCategory: failure.failureCategory,
+                    durationMs: Date.now() - started,
+                };
             }
             const reportsDir = path.join(meta.runDir, "reports");
             const pytestHtmlContent = await readFile(path.join(reportsDir, "backend-test.html"), "utf8");

package/dist/worker/delivery/git-transaction.js

@@ -101,11 +101,12 @@ export async function finalizeGitTask(transaction, outcome) {
     await assertTransactionPosition(transaction.repoRoot, current);
     const currentIgnored = await readIgnoredBaseline(transaction.repoRoot);
     const ignoredBaselineByPath = new Map(current.ignoredBaseline.map((entry) => [entry.path, entry.sha256]));
-    const changedBaselineIgnored = currentIgnored.filter((entry) => ignoredBaselineByPath.has(entry.path) && ignoredBaselineByPath.get(entry.path) !== entry.sha256).map((entry) => entry.path);
-    const missingBaselineIgnored = current.ignoredBaseline.filter((entry) => !currentIgnored.some((candidate) => candidate.path === entry.path)).map((entry) => entry.path);
-    if (changedBaselineIgnored.length > 0 || missingBaselineIgnored.length > 0) {
-        throw new Error(`pre-existing ignored file changed during Git transaction: ${[...changedBaselineIgnored, ...missingBaselineIgnored].join(", ")}`);
-    }
+    const changedBaselineIgnored = currentIgnored
+        .filter((entry) => ignoredBaselineByPath.has(entry.path) && ignoredBaselineByPath.get(entry.path) !== entry.sha256)
+        .map((entry) => entry.path);
+    const missingBaselineIgnored = current.ignoredBaseline
+        .filter((entry) => !currentIgnored.some((candidate) => candidate.path === entry.path))
+        .map((entry) => entry.path);
     const existing = current.checkpoints.find((entry) => entry.workerRunId === outcome.workerRunId);
     if (outcome.status === "reused" || existing) {
         if (!existing)
@@ -123,12 +124,28 @@ export async function finalizeGitTask(transaction, outcome) {
     const newIgnored = currentIgnored.map((entry) => entry.path).filter((entry) => !ignoredBaselineByPath.has(entry));
     if (outcome.status === "succeeded") {
         const allowedPaths = expandAllowedPathsForWorkflow(resolveWorkflow(outcome.taskSpec).workflow, outcome.taskSpec.constraints.allowed_paths);
+        const isAllowedEvidencePath = (entry) => !isSensitivePath(entry) &&
+            !isEphemeralToolCachePath(entry) &&
+            allowedPaths.some((glob) => matchesGlob(entry, glob));
+        // Pre-existing ignored evidence under allowed_paths may be rewritten by
+        // FINAL-VERIFY smoke / verify-final and must force-add into the checkpoint.
+        // Other baseline ignored mutations remain fail-closed (e.g. dist/, secrets).
+        const blockedBaselineChanged = changedBaselineIgnored.filter((entry) => !isAllowedEvidencePath(entry));
+        const blockedBaselineMissing = missingBaselineIgnored.filter((entry) => !isEphemeralToolCachePath(entry) && !isAllowedEvidencePath(entry));
+        if (blockedBaselineChanged.length > 0 || blockedBaselineMissing.length > 0) {
+            throw new Error(`pre-existing ignored file changed during Git transaction: ${[...blockedBaselineChanged, ...blockedBaselineMissing].join(", ")}`);
+        }
         // Evidence under allowed_paths may be gitignored for local convenience (e.g.
         // reports/welcome/** smoke JSON) but must still checkpoint for Task Pool Done.
         // Sensitive / out-of-scope ignored files remain fail-closed; ephemeral tool
         // caches neither block nor force-add.
         const forceAddIgnored = [];
         const blockedIgnored = [];
+        const rewrittenBaselineEvidence = changedBaselineIgnored.filter((entry) => isAllowedEvidencePath(entry));
+        for (const entry of rewrittenBaselineEvidence) {
+            if (!forceAddIgnored.includes(entry))
+                forceAddIgnored.push(entry);
+        }
         for (const entry of newIgnored) {
             if (isEphemeralToolCachePath(entry))
                 continue;
@@ -137,7 +154,8 @@ export async function finalizeGitTask(transaction, outcome) {
                 continue;
             }
             if (allowedPaths.some((glob) => matchesGlob(entry, glob))) {
-                forceAddIgnored.push(entry);
+                if (!forceAddIgnored.includes(entry))
+                    forceAddIgnored.push(entry);
             }
             else {
                 blockedIgnored.push(entry);
@@ -166,7 +184,10 @@ export async function finalizeGitTask(transaction, outcome) {
             await assertTransactionPosition(transaction.repoRoot, { ...current, lastCheckpoint: commit });
             await assertCheckpointMetadata(transaction.repoRoot, commit, outcome.taskSpec, outcome.workerRunId);
             const afterCommitIgnored = await readIgnoredBaseline(transaction.repoRoot);
-            assertIgnoredBaselineUnchanged(current.ignoredBaseline.filter((entry) => !isEphemeralToolCachePath(entry.path)), afterCommitIgnored.filter((entry) => !isEphemeralToolCachePath(entry.path)));
+            // force-add may promote allowed evidence from ignored → tracked; drop those
+            // paths from the post-commit baseline equality check.
+            const forceAdded = new Set(forceAddIgnored);
+            assertIgnoredBaselineUnchanged(current.ignoredBaseline.filter((entry) => !isEphemeralToolCachePath(entry.path) && !forceAdded.has(entry.path)), afterCommitIgnored.filter((entry) => !isEphemeralToolCachePath(entry.path) && !forceAdded.has(entry.path)));
         }
         catch (error) {
             const branch = await git(transaction.repoRoot, ["branch", "--show-current"]).catch(() => "");
@@ -199,6 +220,10 @@ export async function finalizeGitTask(transaction, outcome) {
             changedFiles: checkpointFiles,
         };
     }
+    // Failed / keep-diff paths still fail closed on unexpected baseline ignored mutations.
+    if (changedBaselineIgnored.length > 0 || missingBaselineIgnored.length > 0) {
+        throw new Error(`pre-existing ignored file changed during Git transaction: ${[...changedBaselineIgnored, ...missingBaselineIgnored].join(", ")}`);
+    }
     const artifactDir = path.join(path.dirname(transaction.recordPath), "failures", outcome.workerRunId);
     await captureFailureArtifacts(transaction.repoRoot, artifactDir, changes, newIgnored, outcome);
     if (outcome.keepFailedDiff) {

package/dist/worker/runner/run-ready.js

@@ -304,11 +304,51 @@ export async function runReadyTasks(options) {
             if (result.status !== "succeeded") {
                 await options.onTaskFinalized?.({ status: result.status, taskSpec, workerRunId: result.workerRunId, runRecordPath: result.runRecordPath });
             }
+            const recordErrorMessage = errorMessage(error);
+            // Always project a Pool run fact for record-error so operators can
+            // mark-failed / retry without hand-editing Running state.
+            try {
+                await recordTaskPoolRun({
+                    repoRoot: options.repoRoot,
+                    run: {
+                        schemaVersion: 1,
+                        batchRunId,
+                        workerRunId: result.workerRunId,
+                        taskId: result.businessId,
+                        featureId: taskSpec.feature_id,
+                        status: "run-error",
+                        recordedAt: new Date().toISOString(),
+                        error: recordErrorMessage,
+                        runRecordPath: result.runRecordPath,
+                        ...(result.workflow ? { workflow: result.workflow } : {}),
+                        ...(controllerIdentity ? { controllerIdentity } : {}),
+                        ...(retryOfWorkerRunId ? { retryOfWorkerRunId } : {}),
+                    },
+                });
+                await writeTaskPoolState(options.repoRoot, {
+                    schemaVersion: 2,
+                    featureId: taskSpec.feature_id,
+                    taskId,
+                    status: "Failed",
+                    updatedAt: new Date().toISOString(),
+                    workerRunId: result.workerRunId,
+                    lastRunRecordPath: result.runRecordPath,
+                    failure: {
+                        category: "EnvFailure",
+                        recommendedFollowUpKind: "manual-review",
+                        derivedFollowUpTaskId: `${taskId}-record-error-review`,
+                        source: "report-decision",
+                    },
+                });
+            }
+            catch {
+                // Best-effort; batch still reports record-error.
+            }
             tasks.push({
                 taskId,
                 workerRunId: result.workerRunId,
                 status: "record-error",
-                error: errorMessage(error),
+                error: recordErrorMessage,
                 runRecordPath: result.runRecordPath,
             });
             break;

package/examples/l5-report-demo.html

@@ -0,0 +1,79 @@
+<!doctype html>
+<html lang="zh-CN">
+<head>
+  <meta charset="utf-8">
+  <meta name="viewport" content="width=device-width, initial-scale=1">
+  <title>测试成熟度报告 · L-5</title>
+  <style>
+    :root{--page:#f5f7fb;--surface:#fff;--surface-2:#f8fafd;--ink:#17243b;--muted:#718097;--line:#e4e9f1;--navy:#17365d;--blue:#4775ef;--blue-soft:#edf3ff;--green:#15815c;--green-soft:#e7f7f0;--amber:#c87918;--amber-soft:#fff4e5;--red:#d34661;--red-soft:#fff0f3;--shadow:0 8px 24px rgba(24,54,93,.055)}
+    *{box-sizing:border-box}body{margin:0;background:radial-gradient(circle at 8% 0,#edf4ff 0,transparent 36rem),var(--page);color:var(--ink);font:15px/1.6 Inter,system-ui,-apple-system,BlinkMacSystemFont,"Segoe UI","Microsoft YaHei",sans-serif}
+    main{width:min(1180px,calc(100% - 56px));margin:auto;padding:42px 0 72px}.topbar{display:flex;align-items:flex-start;justify-content:space-between;gap:28px}.eyebrow{color:#71809a;font-size:12px;font-weight:800;letter-spacing:.16em}.title{margin:14px 0 8px;color:#101d34;font-size:clamp(34px,4.2vw,52px);font-weight:800;letter-spacing:-.055em;line-height:1.1}.meta{margin:0;color:var(--muted);font-size:15px}.decision{min-width:236px;padding:15px 18px;border:1px solid #efbdc8;border-radius:18px;background:linear-gradient(135deg,#fff8f9,#fff1f3);color:#c53b58}.decision-top{display:flex;align-items:center;gap:9px;font-size:17px;font-weight:800}.decision-dot{width:11px;height:11px;border-radius:50%;background:var(--red);box-shadow:0 0 0 5px rgba(211,70,97,.1)}.decision small{display:block;margin:7px 0 0 20px;color:#a95e70;font-size:12px}.stats{display:grid;grid-template-columns:repeat(4,1fr);gap:14px;margin-top:32px}.stat{position:relative;min-height:146px;padding:23px 24px;overflow:hidden;background:var(--surface);border:1px solid var(--line);border-radius:17px;box-shadow:var(--shadow)}.stat:before{content:"";position:absolute;top:0;left:0;right:0;height:4px;background:var(--blue)}.stat:nth-child(2):before{background:var(--amber)}.stat:nth-child(3):before{background:#7589d9}.stat:nth-child(4):before{background:var(--green)}.stat-label{color:#748198;font-size:15px;font-weight:650}.stat-value{margin-top:23px;color:#13213a;font-size:38px;font-weight:800;letter-spacing:-.06em;line-height:1}.stat-foot{margin-top:9px;color:var(--muted);font-size:13px}.stat-foot.warn{color:var(--amber)}.stat-foot.good{color:var(--green)}.layout{display:grid;grid-template-columns:1.28fr .86fr;gap:20px;margin-top:20px}.panel{background:var(--surface);border:1px solid var(--line);border-radius:17px;box-shadow:var(--shadow)}.panel-pad{padding:24px}.panel-heading{display:flex;align-items:center;justify-content:space-between;gap:14px;margin-bottom:21px}.panel-heading h2{margin:0;color:var(--navy);font-size:21px;letter-spacing:-.025em}.panel-note{color:var(--muted);font-size:13px}.gate{padding:16px 17px;border:1px solid #e4ebf4;border-radius:13px;background:linear-gradient(135deg,#fbfdff,#f7faff)}.gate+.gate{margin-top:10px}.gate-head{display:flex;align-items:center;gap:10px}.gate-name{flex:1;color:#263c5e;font-size:14px;font-weight:750}.gate-value{font-size:16px;font-weight:800}.gate-value.pass{color:var(--green)}.gate-value.warn{color:var(--amber)}.gate-value.fail{color:var(--red)}.meter{height:7px;margin:11px 0 8px;background:#e7edf5;border-radius:99px;overflow:hidden}.meter i{display:block;height:100%;border-radius:inherit;background:var(--blue)}.meter i.pass{background:var(--green)}.meter i.warn{background:#e9a040}.meter i.fail{background:var(--red)}.gate-foot{display:flex;justify-content:space-between;color:var(--muted);font-size:12px}.badge{display:inline-flex;align-items:center;padding:3px 9px;border-radius:999px;font-size:11px;font-weight:800}.badge.pass{color:var(--green);background:var(--green-soft)}.badge.warn{color:var(--amber);background:var(--amber-soft)}.badge.fail{color:var(--red);background:var(--red-soft)}.cause{padding:16px 0;border-bottom:1px solid var(--line)}.cause:first-of-type{padding-top:0}.cause:last-child{padding-bottom:0;border-bottom:0}.cause-head{display:flex;align-items:center;gap:10px}.cause-label{padding:4px 9px;border-radius:7px;background:var(--red-soft);color:var(--red);font-size:12px;font-weight:800}.cause-label.env{background:var(--amber-soft);color:var(--amber)}.cause-label.none{background:#eef2f7;color:#65738b}.cause-title{flex:1;color:#2a3c5a;font-size:15px;font-weight:700}.cause-count{font-size:18px;font-weight:800}.cause-bar{height:7px;margin-top:12px;background:#edf1f6;border-radius:99px;overflow:hidden}.cause-bar i{display:block;height:100%;width:76%;background:var(--red);border-radius:inherit}.cause-bar i.env{width:48%;background:#e9a040}.cause-bar i.none{width:24%;background:#8491a6}.full{margin-top:20px}.defects{display:grid;grid-template-columns:repeat(2,minmax(0,1fr));gap:12px}.defect{border:1px solid #f0c7d0;border-radius:13px;background:#fffafb;overflow:hidden}.defect summary{display:flex;align-items:center;gap:10px;padding:15px;cursor:pointer;list-style:none}.defect summary::-webkit-details-marker{display:none}.bug-id{color:#c43b58;font:700 12px ui-monospace,SFMono-Regular,Menlo,monospace}.bug-title{flex:1;color:var(--navy);font-size:14px;font-weight:750}.bug-body{padding:0 15px 15px;border-top:1px solid #f3dce1}.bug-fields{display:grid;grid-template-columns:repeat(2,minmax(0,1fr));gap:9px;margin-top:14px}.bug-field{padding:9px 10px;border-radius:8px;background:#fff}.bug-field label{display:block;color:var(--muted);font-size:11px}.bug-field span{display:block;margin-top:2px;color:var(--ink);font-size:12px;font-weight:650}.suggestions{display:grid;grid-template-columns:repeat(3,1fr);gap:12px}.suggestion{padding:17px;border:1px solid #e3eafa;border-radius:13px;background:linear-gradient(135deg,#fbfcff,#f4f7ff)}.priority{display:inline-grid;place-items:center;width:28px;height:24px;border-radius:7px;background:var(--blue-soft);color:var(--blue);font-size:11px;font-weight:850}.suggestion strong{display:block;margin-top:11px;color:var(--navy);font-size:14px}.suggestion span{display:block;margin-top:5px;color:var(--muted);font-size:12px}.details{margin-top:20px}.case{border-top:1px solid var(--line)}details summary{cursor:pointer;list-style:none}details summary::-webkit-details-marker{display:none}.case-summary{display:flex;align-items:center;gap:11px;padding:16px 2px}.case-id{padding:4px 9px;border-radius:6px;background:#edf2f8;color:var(--navy);font:700 12px ui-monospace,SFMono-Regular,Menlo,monospace}.case-title{flex:1;color:#263c5e;font-weight:750}.case-meta{color:var(--muted);font-size:13px}.case-body{display:grid;grid-template-columns:repeat(2,minmax(0,1fr));gap:10px;padding:0 0 18px}.case-box{padding:13px;border:1px solid var(--line);border-radius:10px;background:#fbfcfe}.case-box h3{margin:0 0 5px;color:var(--muted);font-size:12px}.case-box pre{margin:0;white-space:pre-wrap;color:#344054;font:12px/1.6 ui-monospace,SFMono-Regular,Menlo,monospace}.evidence{margin-top:10px;padding-top:9px;border-top:1px solid #e5ebf3}.evidence summary{color:var(--blue);font-size:12px;font-weight:750}.evidence-body{margin-top:7px;color:var(--muted);font-size:12px}.evidence-body code{color:var(--navy);font:11px ui-monospace,SFMono-Regular,Menlo,monospace}.footnote{margin:18px 0 0;color:var(--muted);font-size:12px}@media(max-width:900px){main{width:min(100% - 32px,1180px)}.stats{grid-template-columns:repeat(2,1fr)}.layout{grid-template-columns:1fr}.suggestions{grid-template-columns:1fr}}@media(max-width:560px){main{width:calc(100% - 20px);padding-top:25px}.topbar{display:block}.decision{margin-top:18px}.stats,.defects,.bug-fields,.case-body{grid-template-columns:1fr}.stat{min-height:126px;padding:19px}.stat-value{margin-top:18px;font-size:34px}.panel-pad{padding:19px}.panel-heading{align-items:flex-start;flex-direction:column}.case-meta{display:none}}
+    .topbar{position:relative;overflow:hidden;padding:34px 38px;border:1px solid #294b78;border-radius:26px;background:linear-gradient(125deg,#102849 0%,#173d6d 57%,#245b91 100%);box-shadow:0 18px 40px rgba(16,40,73,.18)}
+    .topbar:before,.topbar:after{content:"";position:absolute;border:1px solid rgba(165,211,255,.18);border-radius:50%;pointer-events:none}.topbar:before{width:370px;height:370px;right:-155px;top:-225px}.topbar:after{width:240px;height:240px;right:55px;bottom:-190px}
+    .topbar>div:first-child{position:relative;z-index:1}.eyebrow{color:#9fc6ee;font-size:11px;letter-spacing:.2em}.title{margin:12px 0 10px;color:#f7fbff;font-size:clamp(33px,4vw,48px);letter-spacing:-.06em}.meta{color:#bdd2e9;font-size:13px}
+    .decision{position:relative;z-index:1;min-width:220px;padding:16px 18px;border:1px solid rgba(255,177,192,.35);background:rgba(255,245,247,.1);box-shadow:0 8px 24px rgba(0,0,0,.08)}.decision-top{color:#fff1f4}.decision small{color:#f3bec8}
+    .stats{gap:12px;margin-top:18px}.stat{min-height:132px;padding:21px 22px;border:1px solid #e1e8f2;border-radius:15px;box-shadow:0 6px 18px rgba(25,53,92,.045)}.stat:before{height:3px}.stat-label{font-size:13px}.stat-value{margin-top:19px;font-size:35px}.stat-foot{font-size:12px}
+    .panel{border-color:#e0e7f0;border-radius:15px;box-shadow:0 7px 20px rgba(25,53,92,.04)}.panel-pad{padding:22px}.panel-heading{margin-bottom:18px}.panel-heading h2{font-size:19px}.panel-heading h2:before{content:"";display:inline-block;width:4px;height:17px;margin:0 9px 0 0;vertical-align:-2px;border-radius:5px;background:var(--blue)}
+    .gate{padding:14px 15px;border-color:#e5ebf3;border-radius:11px;background:#fcfdff}.gate+.gate{margin-top:8px}.gate-name{font-size:13px}.gate-value{font-size:15px}.meter{height:6px;margin:10px 0 7px}.evidence summary{display:flex;align-items:center;gap:6px}.evidence summary:before{content:"+";display:inline-grid;place-items:center;width:15px;height:15px;border:1px solid currentColor;border-radius:50%;font-size:13px;line-height:1}.evidence[open] summary:before{content:"−"}
+    .cause{padding:14px 0}.cause-title{font-size:14px}.cause-count{font-size:16px}.cause-bar{height:5px;margin-top:10px}
+    .full{margin-top:16px}.defects{gap:10px}.defect{border-color:#ecd4d9;border-radius:11px;background:linear-gradient(135deg,#fffdfd,#fff8f9)}.defect summary{padding:14px}.bug-title{font-size:13px}.bug-body{padding:0 14px 14px}.bug-fields{gap:7px;margin-top:12px}.bug-field{padding:8px 9px}.bug-field span{font-size:11px}
+    .suggestions{gap:10px}.suggestion{padding:15px;border-color:#e0e8f5;border-radius:11px}.suggestion strong{margin-top:9px;font-size:13px}.suggestion span{font-size:11px}.priority{width:27px;height:23px}
+    .details{margin-top:16px}.case-summary{padding:14px 2px}.case-title{font-size:13px}.case-box{padding:12px;border-radius:9px}.case-box h3{font-size:11px}.case-box pre{font-size:11px}
+    details:not(.evidence) summary:after{content:"⌄";margin-left:auto;color:#91a0b5;font-size:16px;transition:transform .2s ease}details[open]:not(.evidence) summary:after{transform:rotate(180deg)}
+    .overview{display:grid;grid-template-columns:1fr 1.35fr;gap:12px;margin-top:18px}.overview-chart,.signal-chart{padding:20px 22px;border:1px solid #dfe7f1;border-radius:15px;background:rgba(255,255,255,.82);box-shadow:0 7px 20px rgba(25,53,92,.04)}.chart-heading{display:flex;align-items:flex-start;justify-content:space-between;gap:16px}.chart-heading h2{margin:3px 0 0;color:var(--navy);font-size:17px;letter-spacing:-.02em}.chart-kicker{color:#8796aa;font-size:10px;font-weight:800;letter-spacing:.16em}.chart-caption{color:var(--muted);font-size:11px;white-space:nowrap}.donut-wrap{display:flex;align-items:center;gap:20px;margin-top:15px}.donut{width:116px;height:116px;flex:0 0 116px;transform:rotate(-90deg)}.donut circle{fill:none;stroke-width:12}.donut-track{stroke:#edf1f6}.donut-pass{stroke:var(--green);stroke-dasharray:267.7 289;stroke-linecap:round}.donut-fail{stroke:var(--red);stroke-dasharray:16 289;stroke-dashoffset:-267.7}.donut-center{position:absolute;display:flex;width:116px;height:116px;flex-direction:column;align-items:center;justify-content:center;pointer-events:none}.donut-center strong{color:var(--navy);font-size:20px;letter-spacing:-.05em}.donut-center span{color:var(--muted);font-size:10px}.donut-wrap{position:relative}.outcome-list{display:grid;gap:6px;flex:1}.outcome-list div{display:grid;grid-template-columns:10px 1fr auto;align-items:center;gap:7px;color:var(--muted);font-size:12px}.outcome-list strong{color:var(--ink);font-size:14px}.legend-dot{width:8px;height:8px;border-radius:50%;background:var(--green)}.legend-dot.fail{background:var(--red)}.legend-dot.blocked{background:#aab5c4}.signal-chart{display:flex;flex-direction:column}.signal-row{margin-top:14px}.signal-row>div:first-child{display:flex;justify-content:space-between;align-items:baseline;color:#2b405f;font-size:12px}.signal-row strong{color:var(--navy);font-size:14px}.signal-track{position:relative;height:8px;margin:7px 0 4px;background:#edf1f6;border-radius:99px}.signal-track i{display:block;height:100%;border-radius:inherit;background:linear-gradient(90deg,#7d9bf5,var(--blue))}.signal-track i.good{background:linear-gradient(90deg,#52b88b,var(--green))}.signal-track b{position:absolute;top:-3px;width:2px;height:14px;background:#273b59;border-radius:2px}.signal-row small{color:var(--muted);font-size:10px}
+    @media(max-width:900px){.overview{grid-template-columns:1fr}.donut-wrap{max-width:310px}}
+    @media(max-width:560px){.topbar{padding:28px 23px;border-radius:21px}.stats{gap:9px}.stat{min-height:118px;padding:17px}.stat-value{font-size:31px}.panel-pad{padding:18px}.case-summary{gap:7px}.case-title{min-width:0}.case-summary .case-title{white-space:nowrap;overflow:hidden;text-overflow:ellipsis}}
+  </style>
+</head>
+<body>
+  <main>
+    <header class="topbar"><div><div class="eyebrow">BACKEND TEST · L-5 QUALITY VIEW</div><h1 class="title">测试成熟度报告</h1><p class="meta">my-webapp / Campaign R04 · commit 7d9a2f1 · 2026-07-24 10:42</p></div><div class="decision"><div class="decision-top"><i class="decision-dot"></i>L-5 not ready</div><small>3 个阻断项需要处理</small></div></header>
+    <section class="overview" aria-label="执行概览">
+      <div class="overview-chart">
+        <div class="chart-heading"><div><span class="chart-kicker">RUN OUTCOME</span><h2>本轮执行结果</h2></div><span class="chart-caption">18 个用例</span></div>
+        <div class="donut-wrap">
+          <svg class="donut" viewBox="0 0 120 120" role="img" aria-label="17 个通过，1 个失败，0 个阻塞">
+            <circle class="donut-track" cx="60" cy="60" r="46"></circle>
+            <circle class="donut-pass" cx="60" cy="60" r="46"></circle>
+            <circle class="donut-fail" cx="60" cy="60" r="46"></circle>
+          </svg>
+          <div class="donut-center"><strong>94.4%</strong><span>通过率</span></div>
+          <div class="outcome-list"><div><i class="legend-dot pass"></i><span>通过</span><strong>17</strong></div><div><i class="legend-dot fail"></i><span>失败</span><strong>1</strong></div><div><i class="legend-dot blocked"></i><span>阻塞</span><strong>0</strong></div></div>
+        </div>
+      </div>
+      <div class="signal-chart">
+        <div class="chart-heading"><div><span class="chart-kicker">QUALITY SIGNALS</span><h2>质量信号</h2></div><span class="chart-caption">按 L-5 门槛判定</span></div>
+        <div class="signal-row"><div><span>需求 / AC 覆盖</span><strong>85.7%</strong></div><div class="signal-track"><i style="width:85.7%"></i><b style="left:100%"></b></div><small>12 / 14 · 目标 100%</small></div>
+        <div class="signal-row"><div><span>自动化覆盖率</span><strong>88.9%</strong></div><div class="signal-track"><i style="width:88.9%"></i><b style="left:90%"></b></div><small>16 / 18 · 目标 ≥90%</small></div>
+        <div class="signal-row"><div><span>Line 代码覆盖</span><strong>82.4%</strong></div><div class="signal-track"><i class="good" style="width:82.4%"></i><b style="left:80%"></b></div><small>103 / 125 · 目标 ≥80%</small></div>
+      </div>
+    </section>
+    <section class="layout">
+      <article class="panel panel-pad"><div class="panel-heading"><h2>L-5 门禁摘要</h2><span class="panel-note">阻断项与执行门槛</span></div>
+        <div class="gate"><div class="gate-head"><span class="gate-name">测试通过率</span><strong class="gate-value warn">94.4%</strong><span class="badge warn">未达标</span></div><div class="meter"><i class="warn" style="width:94.4%"></i></div><div class="gate-foot"><span>17 / 18 个用例</span><span>目标 100%</span></div><details class="evidence"><summary>查看计算依据</summary><div class="evidence-body">失败 1 条，错误 0 条。证据：<code>reports/backend-test-facts.md · pytest-exit=1</code></div></details></div>
+        <div class="gate"><div class="gate-head"><span class="gate-name">Critical 风险</span><strong class="gate-value fail">1</strong><span class="badge fail">阻断</span></div><div class="meter"><i class="fail" style="width:35%"></i></div><div class="gate-foot"><span>1 个阻断项</span><span>目标 0</span></div><details class="evidence"><summary>查看阻断原因</summary><div class="evidence-body">R-01：手机号注册校验返回错误业务结果。关联：<code>AC-001 · TC-017 · ProductBug</code></div></details></div>
+      </article>
+      <article class="panel panel-pad"><div class="panel-heading"><h2>失败诊断</h2><span class="panel-note">1 个失败 · 1 个错误</span></div>
+        <div class="cause"><div class="cause-head"><span class="cause-label">ProductBug</span><span class="cause-title">手机号注册校验</span><strong class="cause-count">1</strong></div><div class="cause-bar"><i></i></div><details class="evidence"><summary>查看诊断详情</summary><div class="evidence-body">期望 HTTP 400，实际 HTTP 200。关联：TC-017 / AC-001。</div></details></div>
+        <div class="cause"><div class="cause-head"><span class="cause-label env">EnvFailure</span><span class="cause-title">依赖服务连接</span><strong class="cause-count">1</strong></div><div class="cause-bar"><i class="env"></i></div><details class="evidence"><summary>查看环境详情</summary><div class="evidence-body">GET /health/dependency 超时 5000ms，未形成有效业务断言。</div></details></div>
+        <div class="cause"><div class="cause-head"><span class="cause-label none">Unknown</span><span class="cause-title">缺少有效证据</span><strong class="cause-count">0</strong></div><div class="cause-bar"><i class="none"></i></div><details class="evidence"><summary>查看判定规则</summary><div class="evidence-body">当前没有无法分类的失败记录。</div></details></div>
+      </article>
+    </section>
+    <section class="panel panel-pad full"><div class="panel-heading"><h2>缺陷记录</h2><span class="panel-note">2 条 · 已关联用例与证据</span></div><div class="defects">
+      <details class="defect"><summary><span class="bug-id">BUG-20260724-017</span><span class="bug-title">手机号注册校验未拦截非法格式</span><span class="badge fail">未修复</span></summary><div class="bug-body"><div class="bug-fields"><div class="bug-field"><label>级别</label><span>High · ProductBug</span></div><div class="bug-field"><label>模块</label><span>用户注册 / 手机号校验</span></div><div class="bug-field"><label>关联 AC / Case</label><span>AC-001 · TC-017</span></div><div class="bug-field"><label>发现版本</label><span>Campaign R04 · 7d9a2f1</span></div></div><details class="evidence"><summary>查看缺陷证据</summary><div class="evidence-body">期望 HTTP 400，实际 HTTP 200。<br>本地记录：<code>defects/BUG-20260724-017.md</code><br>关联证据：<code>reports/evidence/TC-017-response.json</code></div></details></div></details>
+      <details class="defect"><summary><span class="bug-id">BUG-20260724-018</span><span class="bug-title">依赖服务健康检查超时</span><span class="badge warn">待确认</span></summary><div class="bug-body"><div class="bug-fields"><div class="bug-field"><label>级别</label><span>Medium · EnvFailure</span></div><div class="bug-field"><label>模块</label><span>审批服务依赖检查</span></div><div class="bug-field"><label>关联 AC / Case</label><span>AC-004 · TC-018</span></div><div class="bug-field"><label>发现版本</label><span>Campaign R04 · 7d9a2f1</span></div></div><details class="evidence"><summary>查看环境证据</summary><div class="evidence-body">GET /health/dependency 超时 5000ms。<br>本地记录：<code>defects/BUG-20260724-018.md</code><br>建议：补充服务可用性和网络诊断信息。</div></details></div></details>
+    </div></section>
+    <section class="panel panel-pad full"><div class="panel-heading"><h2>下一轮回归建议</h2><span class="panel-note">3 条行动项</span></div><div class="suggestions">
+      <div class="suggestion"><span class="priority">P0</span><strong>修复注册校验后重跑 TC-017</strong><span>核对 AC-001、HTTP 400 与错误码。</span></div>
+      <div class="suggestion"><span class="priority">P1</span><strong>补充依赖服务健康检查</strong><span>记录 URL、超时配置、错误类型和时间戳。</span></div>
+      <div class="suggestion"><span class="priority">P2</span><strong>补齐缺失 AC 自动化映射</strong><span>覆盖 AC-013、AC-014，提升追踪完整度。</span></div>
+    </div></section>
+    <section class="panel panel-pad full details"><div class="panel-heading"><h2>用例执行明细</h2><span class="panel-note">共 3 条 · 点击展开</span></div>
+      <details class="case"><summary class="case-summary"><span class="case-id">TC-017</span><span class="case-title">手机号注册校验</span><span class="case-meta">失败 · 1.24s</span><span class="badge fail">FAIL</span></summary><div class="case-body"><div class="case-box"><h3>失败原因</h3><div>ProductBug · 返回 200，未展示预期格式错误提示。</div></div><div class="case-box"><h3>自动化脚本</h3><pre>testcase/test_register.py::test_TC_017</pre></div><div class="case-box"><h3>请求参数</h3><pre>POST /api/register · phone=138****8000</pre></div><div class="case-box"><h3>响应结果</h3><pre>200 OK · {"success":true,"message":"注册成功"}</pre></div></div></details>
+      <details class="case"><summary class="case-summary"><span class="case-id">TC-018</span><span class="case-title">依赖服务连接检查</span><span class="case-meta">错误 · 0.82s</span><span class="badge warn">ERROR</span></summary><div class="case-body"><div class="case-box"><h3>失败原因</h3><div>EnvFailure · 依赖服务连接超时，未形成有效业务断言。</div></div><div class="case-box"><h3>自动化脚本</h3><pre>testcase/test_dependency.py::test_TC_018</pre></div><div class="case-box"><h3>请求参数</h3><pre>GET /health/dependency · timeout=5000ms</pre></div><div class="case-box"><h3>响应结果</h3><pre>Unavailable · connection timed out</pre></div></div></details>
+      <details class="case"><summary class="case-summary"><span class="case-id">TC-019</span><span class="case-title">正常流程提交</span><span class="case-meta">通过 · 0.31s</span><span class="badge pass">PASS</span></summary><div class="case-body"><div class="case-box"><h3>验收场景</h3><div>提交合法业务数据后返回成功状态，并生成业务编号。</div></div><div class="case-box"><h3>自动化脚本</h3><pre>testcase/test_flow.py::test_TC_019</pre></div><div class="case-box"><h3>请求参数</h3><pre>POST /api/flows · {"name":"approval-flow"}</pre></div><div class="case-box"><h3>响应结果</h3><pre>201 Created · {"flowId":"flow-001"}</pre></div></div></details>
+    </section>
+    <p class="footnote">Demo 数据仅用于展示页面结构；实际报告以同一次测试执行产生的 HTML、Markdown 和 facts 为准。</p>
+  </main>
+</body>
+</html>

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@tea-agent/loop-agent",
-  "version": "0.24.7",
+  "version": "0.24.9",
   "type": "module",
   "bin": {
     "loop-agent": "bin/loop-agent.js",