npm - @tea-agent/loop-agent - Versions diffs - 0.22.0 → 0.24.0 - Mend

@tea-agent/loop-agent 0.22.0 → 0.24.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (103) hide show

package/AGENTS.md +42 -108
package/CHANGELOG.md +85 -0
package/README.md +8 -5
package/bin/agent-worker.js +0 -0
package/dist/application/context-usage/skill-resolution-stats.js +263 -0
package/dist/application/dag/generate-task-dag.js +17 -3
package/dist/cli/command-definitions.js +8 -7
package/dist/cli/program.js +17 -15
package/dist/commands/doctor.js +269 -18
package/dist/commands/init.js +101 -86
package/dist/commands/stats.js +40 -11
package/dist/executors/shell-executor.js +20 -7
package/dist/shared/operator/capabilities.js +486 -3
package/dist/worker/console/app-data.js +6 -0
package/dist/worker/console/chat/artifact-card.js +23 -0
package/dist/worker/console/chat/chat-event-store.js +495 -0
package/dist/worker/console/chat/chat-ui-policy.js +25 -0
package/dist/worker/console/chat/composer-draft-store.js +45 -0
package/dist/worker/console/chat/context-panel.js +54 -0
package/dist/worker/console/chat/contract-apply-receipt-store.js +174 -0
package/dist/worker/console/chat/explore-tools.js +299 -0
package/dist/worker/console/chat/human-gate-card.js +37 -0
package/dist/worker/console/chat/instruction-skills.js +217 -0
package/dist/worker/console/chat/interview-adapter.js +136 -0
package/dist/worker/console/chat/model-resolver.js +106 -0
package/dist/worker/console/chat/operation-card.js +23 -0
package/dist/worker/console/chat/pi-console-config.js +158 -0
package/dist/worker/console/chat/pi-runtime.js +1143 -0
package/dist/worker/console/chat/repo-browser.js +140 -0
package/dist/worker/console/chat/repo-walk.js +116 -0
package/dist/worker/console/chat/resource-loader.js +67 -0
package/dist/worker/console/chat/routes.js +1646 -0
package/dist/worker/console/chat/runtime-context.js +24 -0
package/dist/worker/console/chat/runtime-selection.js +37 -0
package/dist/worker/console/chat/session-store.js +437 -0
package/dist/worker/console/chat/shortcuts.js +15 -0
package/dist/worker/console/chat/tool-adapter.js +125 -0
package/dist/worker/console/chat/tools.js +195 -0
package/dist/worker/console/chat/usage.js +37 -0
package/dist/worker/console/chat/workspace-landing.js +56 -0
package/dist/worker/console/dag-confirmation.js +42 -8
package/dist/worker/console/human-gate-token.js +130 -0
package/dist/worker/console/mutation-gate-receipt-store.js +184 -0
package/dist/worker/console/operation-runner.js +6 -2
package/dist/worker/console/operation-sse.js +26 -0
package/dist/worker/console/operator-actions.js +420 -7
package/dist/worker/console/server.js +68 -1
package/dist/worker/console/static/assets/index-BTbrEHnO.css +1 -0
package/dist/worker/console/static/assets/index-D9qLevoP.js +27 -0
package/dist/worker/console/static/index.html +2 -2
package/dist/worker/feature/profile-schema.js +1 -1
package/dist/workflows/dag/backend-test-markdown-workflow.js +9 -5
package/dist/workflows/dag/backend-test-result-contract.js +229 -0
package/dist/workflows/dag/frontend-lint-baseline.js +4 -4
package/dist/workflows/dag/init-hybrid.js +71 -22
package/dist/workflows/dag/node-execution.js +38 -1
package/dist/workflows/dag/output-protocol.js +89 -0
package/dist/workflows/dag/prompt.js +35 -1
package/dist/workflows/dag/recovery-recommendation.js +45 -0
package/dist/workflows/dag/report.js +28 -1
package/dist/workflows/dag/rerun-task.js +1 -1
package/dist/workflows/dag/scheduler.js +9 -0
package/dist/workflows/dag/types.js +12 -0
package/dist/workflows/dag/validate.js +55 -0
package/docs/README.md +73 -156
package/docs/architecture/README.md +7 -6
package/docs/architecture/dag-execution.md +2 -2
package/docs/architecture/evolution.md +16 -14
package/docs/architecture/system-overview.md +1 -1
package/docs/architecture/worker-and-feature.md +3 -3
package/docs/governance/README.md +15 -0
package/docs/{harness-methodology-debugging.md → governance/harness-methodology-debugging.md} +27 -3
package/docs/init-surface.manifest.json +22 -4
package/docs/operations/README.md +12 -0
package/docs/{local-development-environment.md → operations/local-development-environment.md} +1 -1
package/docs/skills/vetted-skill-registry.md +23 -3
package/docs/templates/README.md +55 -0
package/docs/templates/backend-test-dag.json +2 -2
package/docs/templates/evaluation/agents-map-slim-v1.candidate.json +9 -0
package/docs/templates/evaluation/agents-map-slim-v1.md +87 -0
package/docs/templates/evaluation/agents-map-verbose-v0.candidate.json +9 -0
package/docs/templates/evaluation/agents-map-verbose-v0.md +153 -0
package/docs/templates/hybrid-dag.json +1 -1
package/docs/templates/progress-log.md +9 -2
package/harness.json +4 -4
package/package.json +5 -5
package/scripts/kb-bootstrap-init-skeleton.sh +2 -2
package/skills/agent-worker/SKILL.md +1 -1
package/skills/grill-with-docs/SKILL.md +44 -52
package/skills/grill-with-docs/adr-format.md +37 -26
package/skills/grill-with-docs/context-format.md +18 -26
package/skills/loop-agent/SKILL.md +28 -112
package/skills/loop-agent/references/command-reference.md +9 -3
package/skills/loop-agent/references/harness-policy.md +3 -3
package/skills/loop-agent/references/hybrid-dag.md +1 -1
package/skills/loop-agent/references/orchestrator-and-interventions.md +1 -1
package/skills/loop-agent/references/task-workflow.md +2 -0
package/skills/systematic-debugging/SKILL.md +20 -4
package/skills/test-driven-development/SKILL.md +10 -3
package/dist/worker/console/static/assets/index-CUDke82y.js +0 -18
package/dist/worker/console/static/assets/index-wSEksVSO.css +0 -1
/package/docs/{harness-methodology-tdd.md → governance/harness-methodology-tdd.md} +0 -0
/package/docs/{harness-methodology-verification.md → governance/harness-methodology-verification.md} +0 -0

package/dist/worker/console/static/index.html CHANGED Viewed

@@ -4,8 +4,8 @@
     <meta charset="UTF-8" />
     <meta name="viewport" content="width=device-width, initial-scale=1.0" />
     <title>Loop 操作台 · Operator Console</title>
-    <script type="module" crossorigin src="/assets/index-CUDke82y.js"></script>
-    <link rel="stylesheet" crossorigin href="/assets/index-wSEksVSO.css">
+    <script type="module" crossorigin src="/assets/index-D9qLevoP.js"></script>
+    <link rel="stylesheet" crossorigin href="/assets/index-BTbrEHnO.css">
   </head>
   <body>
     <div id="root"></div>

package/dist/worker/feature/profile-schema.js CHANGED Viewed

@@ -5,7 +5,7 @@ import { z } from "zod";
  * `fullstack-v1` is NOT a new DAG or taskKind. It is `agent-worker`'s
  * deterministic structural validation + read-model projection policy over a
  * Feature Packet. See
- * `docs/design/agent-worker-fullstack-workflow-integration.md` §9.
+ * `docs/design/active/agent-worker-fullstack-workflow-integration.md` §9.
  *
  * A Feature Packet declares its profile via an optional `feature.yaml` next to
  * `acceptance.yaml`. A packet without `feature.yaml` (or one that does not set

package/dist/workflows/dag/backend-test-markdown-workflow.js CHANGED Viewed

@@ -97,13 +97,19 @@ export async function inspectBackendTestEnvironment(input) {
         }
         fixtures.push(...extractFixtures(content));
     }
-    const htmlRenderer = /--html\b/.test(input.pytestHelp)
+    const hasPytestHtml = /--html\b/.test(input.pytestHelp);
+    const htmlRenderer = hasPytestHtml
         ? "pytest-html --self-contained-html"
-        : "loop-agent built-in JUnit-to-HTML";
+        : "missing pytest-html (required: pip install pytest-html)";
     const warnings = [
         ...(pytestConfigs.length === 0
             ? ["No explicit pytest configuration file was found."]
             : []),
+        ...(!hasPytestHtml
+            ? [
+                "pytest-html is not installed; the backend-test pipeline requires it to render the self-contained HTML report. Install with: pip install pytest-html",
+            ]
+            : []),
         ...(conftestFiles.length === 0
             ? ["No bounded conftest.py candidate was found."]
             : []),
@@ -843,7 +849,7 @@ export function renderBackendTestFacts(input) {
         `| ${input.pytestExitCode === 0 ? "通过" : "未通过"} | ${input.parsed.tests} | ${input.parsed.passed} | ${input.parsed.failed} | ${input.parsed.errors} | ${input.parsed.skipped} | ${(passRate * 100).toFixed(2)}% | ${formatDuration(input.parsed.durationMs)} |`,
         "",
         `- Pytest exit code: ${input.pytestExitCode}`,
-        "- JUnit valid: yes",
+        "- Pytest-html report: valid",
         "- HTML valid: yes",
         "",
         "## 质量校验",
@@ -872,8 +878,6 @@ export function renderBackendTestFacts(input) {
         "",
         "## 证据与完整校验原文",
         "",
-        `- JUnit: ${input.junitRelativePath}`,
-        `- JUnit SHA-256: ${createHash("sha256").update(input.junitContent).digest("hex")}`,
         `- HTML: ${input.htmlRelativePath}`,
         `- HTML SHA-256: ${createHash("sha256").update(input.htmlContent).digest("hex")}`,
         "",

package/dist/workflows/dag/backend-test-result-contract.js CHANGED Viewed

@@ -354,6 +354,235 @@ export function parseJunitXml(xml) {
         failures: failures.slice(0, MAX_FAILURES),
     };
 }
+const HTML_ENTITY_MAP = {
+    "&amp;": "&",
+    "&lt;": "<",
+    "&gt;": ">",
+    "&quot;": '"',
+    "&#34;": '"',
+    "&#39;": "'",
+    "&apos;": "'",
+};
+function decodeHtmlEntities(value) {
+    return value.replace(/&(?:amp|lt|gt|quot|#34|#39|apos);/g, (entity) => HTML_ENTITY_MAP[entity] ?? entity);
+}
+function parseDurationLabelMs(raw) {
+    if (!raw)
+        return undefined;
+    const value = Number.parseFloat(raw);
+    if (!Number.isFinite(value))
+        return undefined;
+    if (/\bms\b/i.test(raw))
+        return Math.round(value);
+    if (/\bsec|s\b/i.test(raw))
+        return Math.round(value * 1000);
+    // pytest-html commonly emits bare milliseconds without a unit suffix.
+    return Math.round(value);
+}
+/**
+ * Extract the pytest-html 4.x JSON island from a self-contained HTML report.
+ *
+ * pytest-html 4.x embeds the full report payload as an HTML-entity-escaped
+ * JSON string in `<div id="data-container" data-jsonblob="...">`. This helper
+ * locates that attribute, decodes entities and parses the JSON. Fail-closed on
+ * a non-pytest-html document so the pipeline never silently drops test facts.
+ */
+export function extractPytestHtmlReportData(html) {
+    const match = html.match(/id=["']data-container["'][^>]*data-jsonblob=["']([^"']*)["']/i);
+    if (!match || !match[1]) {
+        throw new Error("invalid pytest-html report: missing data-container data-jsonblob island");
+    }
+    const decoded = decodeHtmlEntities(match[1]);
+    try {
+        return JSON.parse(decoded);
+    }
+    catch (error) {
+        throw new Error(`invalid pytest-html report: data-jsonblob is not valid JSON (${error instanceof Error ? error.message : String(error)})`);
+    }
+}
+/**
+ * Parse a self-contained pytest-html 4.x report into the same ParsedJunit
+ * shape used by the HTML/facts renderers, so the downstream rendering path
+ * stays uniform regardless of the evidence source.
+ *
+ * Unlike JUnit, pytest-html stores per-test captured stdout in a `log` field
+ * (with section markers like "Captured stdout call"). We surface that as the
+ * case `stdout` so HTTP_REQUEST/HTTP_RESPONSE lines remain visible in the
+ * per-case cards.
+ */
+export function parsePytestHtmlReport(html) {
+    const data = extractPytestHtmlReportData(html);
+    const entries = Object.entries(data.tests ?? {});
+    if (entries.length === 0) {
+        throw new Error("invalid pytest-html report: no test entries in data-jsonblob");
+    }
+    const cases = [];
+    const failures = [];
+    let passed = 0;
+    let failed = 0;
+    let errors = 0;
+    let skipped = 0;
+    for (const [nodeId, records] of entries) {
+        const record = records[0];
+        if (!record)
+            continue;
+        const rawResult = (record.result ?? "").toLowerCase();
+        const testId = record.testId ?? nodeId;
+        const { classname, name } = splitPytestNodeId(testId);
+        const durationMs = parseDurationLabelMs(record.duration);
+        const capturedLog = record.log ?? "";
+        // pytest-html collapses captured stdout/stderr into a single `log` field
+        // annotated with section markers. Preserve the whole log as stdout so the
+        // HTTP_REQUEST/HTTP_RESPONSE lines stay reachable for the per-case card.
+        const stdout = splitPytestHtmlLogSections(capturedLog).stdout || undefined;
+        const stderr = splitPytestHtmlLogSections(capturedLog).stderr || undefined;
+        if (rawResult === "passed") {
+            passed += 1;
+            cases.push({
+                classname,
+                name,
+                durationMs,
+                status: "passed",
+                ...(stdout ? { stdout } : {}),
+                ...(stderr ? { stderr } : {}),
+            });
+        }
+        else if (rawResult === "failed") {
+            failed += 1;
+            const message = extractPytestHtmlFailureMessage(capturedLog) || "failure";
+            const summary = truncate(message);
+            failures.push({ classname, name, message: summary, kind: "failure" });
+            cases.push({
+                classname,
+                name,
+                durationMs,
+                status: "failure",
+                message: summary,
+                details: capturedLog || summary,
+                ...(stdout ? { stdout } : {}),
+                ...(stderr ? { stderr } : {}),
+            });
+        }
+        else if (rawResult === "error") {
+            errors += 1;
+            const message = extractPytestHtmlFailureMessage(capturedLog) || "error";
+            const summary = truncate(message);
+            failures.push({ classname, name, message: summary, kind: "error" });
+            cases.push({
+                classname,
+                name,
+                durationMs,
+                status: "error",
+                message: summary,
+                details: capturedLog || summary,
+                ...(stdout ? { stdout } : {}),
+                ...(stderr ? { stderr } : {}),
+            });
+        }
+        else if (rawResult === "skipped" || rawResult === "xfailed") {
+            skipped += 1;
+            cases.push({
+                classname,
+                name,
+                durationMs,
+                status: "skipped",
+                ...(stdout ? { stdout } : {}),
+                ...(stderr ? { stderr } : {}),
+            });
+        }
+        else {
+            // Unknown outcome label: treat as error to stay fail-safe.
+            errors += 1;
+            const message = `unexpected pytest-html result label: ${record.result ?? "(empty)"}`;
+            const summary = truncate(message);
+            failures.push({ classname, name, message: summary, kind: "error" });
+            cases.push({
+                classname,
+                name,
+                durationMs,
+                status: "error",
+                message: summary,
+                details: capturedLog || summary,
+                ...(stdout ? { stdout } : {}),
+                ...(stderr ? { stderr } : {}),
+            });
+        }
+    }
+    const tests = cases.length;
+    return {
+        tests,
+        passed,
+        failed,
+        errors,
+        skipped,
+        durationMs: undefined,
+        cases,
+        failures: failures.slice(0, MAX_FAILURES),
+    };
+}
+function splitPytestNodeId(testId) {
+    // pytest-html uses node ids like "path/to/test_x.py::TestClass::test_name".
+    const sepIndex = testId.lastIndexOf("::");
+    if (sepIndex < 0) {
+        return { classname: "unknown", name: testId };
+    }
+    const prefix = testId.slice(0, sepIndex);
+    const name = testId.slice(sepIndex + 2);
+    const lastParen = name.indexOf("(");
+    const cleanName = lastParen >= 0 ? name.slice(0, lastParen) : name;
+    const moduleSep = prefix.lastIndexOf("::");
+    const filePart = moduleSep >= 0 ? prefix.slice(moduleSep + 2) : prefix;
+    const moduleWithoutExt = filePart.replace(/\.py$/i, "");
+    return { classname: moduleWithoutExt.replace(/\//g, ".") || "unknown", name: cleanName };
+}
+function splitPytestHtmlLogSections(log) {
+    if (!log)
+        return { stdout: "", stderr: "" };
+    // pytest-html interleaves captured stdout/stderr with section markers like
+    // "----------------------------- Captured stdout call -----------------------------".
+    // String.split includes capture-group matches in the result array, so the
+    // layout is [beforeText, label, body, label, body, ...]. Iterate label/body
+    // pairs starting at index 1.
+    const sections = log.split(/^-+ Captured (stdout|stderr|log|call|setup|teardown)(?: call| setup| teardown)? -+$/m);
+    let stdout = "";
+    let stderr = "";
+    for (let index = 1; index < sections.length - 1; index += 2) {
+        const label = sections[index]?.toLowerCase() ?? "";
+        const body = sections[index + 1] ?? "";
+        if (label.includes("stderr")) {
+            stderr = `${stderr}${body}`.trim();
+        }
+        else {
+            stdout = `${stdout}${body}`.trim();
+        }
+    }
+    if (!stdout && !stderr) {
+        // No recognizable section markers: treat the whole log as stdout so HTTP
+        // request/response lines remain visible.
+        stdout = log.trim();
+    }
+    return { stdout, stderr };
+}
+function extractPytestHtmlFailureMessage(log) {
+    if (!log)
+        return "";
+    // pytest-html failure logs contain assertion lines prefixed with "E   " and a
+    // trailing location line like "test_x.py:N: AssertionError". Prefer the
+    // explicit AssertionError/Error line; fall back to the last "E   " line.
+    const assertionLine = log
+        .split(/\r?\n/)
+        .map((line) => line.trim())
+        .find((line) => /:\s*AssertionError/.test(line));
+    if (assertionLine)
+        return assertionLine.replace(/^.*?:\s*/, "");
+    const eLines = log
+        .split(/\r?\n/)
+        .filter((line) => /^E\s+/.test(line))
+        .map((line) => line.replace(/^E\s+/, "").trim());
+    if (eLines.length > 0)
+        return eLines[eLines.length - 1];
+    return "";
+}
 export function deriveBackendTestResult(input) {
     assertNoSecrets("commandSummary", input.commandSummary);
     const emptyJunitMeta = {

package/dist/workflows/dag/frontend-lint-baseline.js CHANGED Viewed

@@ -87,14 +87,14 @@ function stripAnsi(value) {
 function normalizeRepoPath(workspaceRoot, candidate) {
     const clean = stripAnsi(candidate).trim().replace(/^file:\/\//, "");
     const windowsAbsolute = /^[A-Za-z]:[\\/]/.test(clean);
-    const windowsRoot = /^[A-Za-z]:[\\/]/.test(workspaceRoot);
+    const posixAbsolute = path.posix.isAbsolute(clean);
     let relative;
-    if (windowsAbsolute || windowsRoot) {
-        if (!windowsAbsolute || !windowsRoot)
+    if (windowsAbsolute) {
+        if (!/^[A-Za-z]:[\\/]/.test(workspaceRoot))
             return null;
         relative = path.win32.relative(workspaceRoot, clean);
     }
-    else if (path.posix.isAbsolute(clean)) {
+    else if (posixAbsolute) {
         relative = path.posix.relative(workspaceRoot.replaceAll("\\", "/"), clean.replaceAll("\\", "/"));
     }
     else {

package/dist/workflows/dag/init-hybrid.js CHANGED Viewed

@@ -18,7 +18,8 @@ import { discoverProjectGovernancePresence } from "./project-governance-context.
 import { getTaskPaths, loadTaskConfig } from "../../task/runtime.js";
 import { materializeTaskReferenceDocs } from "../../task/source-references.js";
 import { observeTaskContract } from "../../task/contract/observe.js";
-import { resolveVerifyPreset } from "../../executors/shell-verification.js";
+import { dagHasWriterExecution } from "./task-contract-binding.js";
+import { DEFAULT_VERIFY_TIMEOUT_MS, resolveVerifyPreset, } from "../../executors/shell-verification.js";
 import { resolveExecutorModelMatrices } from "../../executors/model-routing.js";
 import { normalizeTaskRequirementText, resolveTaskDagTemplateSelection, } from "./task-demand-routing.js";
 import { BACKEND_TEST_EXECUTION_DEFAULT_TEST_ROOT, buildBackendTestExecutionPreflightShellSnippet, } from "./backend-test-execution-contract.js";
@@ -556,7 +557,7 @@ export function hasApiDependency(sources) {
 /**
  * Resolve frontend Mock mode from capability seed, task config, and interface dependency analysis.
  *
- * Decision matrix (from docs/design/frontend-mock-data-workflow.md):
+ * Decision matrix (from docs/design/active/frontend-mock-data-workflow.md):
  *
  * | 接口/异步数据依赖 | 既有 Mock 服务 | policy | 结果 |
  * |---|---|---|---|
@@ -871,21 +872,29 @@ function buildExplicitFrontendVerifyCommands(taskConfig, repoRoot) {
 function verifyCommandKey(command) {
     return `${command.cwd}\0${command.args.join("\0")}`;
 }
-function resolveDagVerifyStrategy(taskConfig) {
+function resolveDagVerifyStrategy(taskConfig, defaultIntermediateQuotaWhenFull = "full") {
+    const explicitIntermediateQuota = taskConfig.dagVerifyStrategy?.intermediateQuota;
     return {
-        intermediateQuota: taskConfig.dagVerifyStrategy?.intermediateQuota ?? taskConfig.verifyQuota,
+        intermediateQuota: explicitIntermediateQuota ??
+            (taskConfig.verifyQuota === "full"
+                ? defaultIntermediateQuotaWhenFull
+                : taskConfig.verifyQuota),
         finalQuota: "full",
         focusedCommandSource: taskConfig.dagVerifyStrategy?.focusedCommandSource ?? "adapter",
     };
 }
 function buildVerifyEvidence(input) {
+    const selectedCommands = input.commands && input.commands.length > 0 ? input.commands : undefined;
+    const commandCount = selectedCommands?.length ?? input.fallbackCommands.length;
     return {
         phase: input.phase,
         quota: input.quota,
         commandSource: input.commandSource,
-        commandCount: input.commands?.length ?? input.fallbackCommands.length,
-        commandLabels: input.commands?.map((command) => command.label) ?? input.fallbackCommands,
+        commandCount,
+        commandLabels: selectedCommands?.map((command) => command.label) ?? input.fallbackCommands,
         commandTexts: input.commandTexts ?? [],
+        commandTimeoutMs: input.commandTimeoutMs,
+        totalTimeoutBudgetMs: commandCount * input.commandTimeoutMs,
         finalFullRequired: input.finalFullRequired,
     };
 }
@@ -1369,9 +1378,10 @@ export function buildStandardHybridDagFromTask(sources) {
                         commands: finalVerifyCommands,
                         fallbackCommands: [],
                         finalFullRequired: true,
+                        commandTimeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
                     }),
                     cwd: ".",
-                    timeoutMs: 300000,
+                    timeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
                 },
             },
         ]
@@ -1677,9 +1687,10 @@ function buildFrontendMockVerifyNode(sources, implementId, readOnlyPaths, forbid
                 commands: verifyCommands.length > 0 ? verifyCommands : undefined,
                 fallbackCommands: [],
                 commandTexts: commands,
+                commandTimeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
             }),
             cwd: ".",
-            timeoutMs: 300000,
+            timeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
         },
     };
 }
@@ -2017,6 +2028,7 @@ async function buildFrontendHybridDagFromTask(sources) {
         commands: partitionedStaticVerifyCommands.static.commands,
         fallbackCommands: staticFallbackCommands,
         commandTexts: staticShellCommands,
+        commandTimeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
     });
     const lintVerifyEvidence = lintShellCommands.length > 0
         ? buildVerifyEvidence({
@@ -2026,6 +2038,7 @@ async function buildFrontendHybridDagFromTask(sources) {
             commands: partitionedStaticVerifyCommands.lint.commands,
             fallbackCommands: [],
             commandTexts: lintShellCommands,
+            commandTimeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
         })
         : undefined;
     const behaviorVerifyEvidence = buildVerifyEvidence({
@@ -2036,6 +2049,7 @@ async function buildFrontendHybridDagFromTask(sources) {
         fallbackCommands: behaviorFallbackCommands,
         commandTexts: behaviorShellCommands,
         finalFullRequired: true,
+        commandTimeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
     });
     const mockVerifyTemplate = mockMode === "required" && hasMockVerifyCommands
         ? buildFrontendMockVerifyNode(frontendSources, implementId, readOnlyPaths, forbiddenPaths)
@@ -2273,7 +2287,7 @@ async function buildFrontendHybridDagFromTask(sources) {
                                 lintEvidence: lintVerifyEvidence,
                             },
                             cwd: ".",
-                            timeoutMs: 300000,
+                            timeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
                         },
                     },
                 ]
@@ -2345,7 +2359,7 @@ async function buildFrontendHybridDagFromTask(sources) {
                         mode: "initial",
                     },
                     cwd: ".",
-                    timeoutMs: 300000,
+                    timeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
                 },
             },
             {
@@ -2406,7 +2420,7 @@ async function buildFrontendHybridDagFromTask(sources) {
                         mode: "repair",
                     },
                     cwd: ".",
-                    timeoutMs: 300000,
+                    timeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
                 },
             },
             {
@@ -3122,9 +3136,10 @@ function buildExecuteBackendPytestNode(sources, options = {}) {
                 commandSource: "inline",
                 fallbackCommands: [pytestCommand],
                 finalFullRequired: true,
+                commandTimeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
             }),
             cwd: ".",
-            timeoutMs: 300000,
+            timeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
         },
     };
 }
@@ -3241,6 +3256,7 @@ function buildBackendTestOutcomeGateNode(sources) {
                 commandSource: "inline",
                 fallbackCommands: [gateCommand],
                 finalFullRequired: true,
+                commandTimeoutMs: 60_000,
             }),
             cwd: ".",
             timeoutMs: 60000,
@@ -3333,10 +3349,11 @@ async function buildBackendTestHybridDag(sources) {
         executor: "pi", toolProfile: "write", complexity: "HIGH", writePolicy: "exclusive",
         writeSet: ["testcase/**/test_*.py", "testcase/**/helpers/**", "testcase/**/factories/**"],
         allowedPaths: Array.from(new Set([...ro, "testcase/**"])), forbiddenPaths: forbidden,
-        outputContract: "Convert every final automatable Markdown case into pytest assets whose actual test function region contains the exact Case ID, preferably in the function name or docstring; no JSON and no pytest execution.",
+        outputContract: "Convert every final automatable Markdown case into pytest assets whose actual test function region contains the exact Case ID, preferably in the function name or docstring. Each testcase/md/<module>.md (excluding README.md) maps one-to-one to testcase/test_<module>.py; never merge or split modules. No JSON and no pytest execution.",
         subtask_prompt: [
             "Convert testcase/md/** to pytest using upstream environment and advisory validation evidence plus only bounded pytest config/conftest. A FAIL advisory report does not authorize inventing missing behavior; use the final Markdown facts that are present.",
             "Ensure every final Markdown Case ID appears in at least one real pytest test function or pytest test class method region, preferably as `test_BE_<MODULE>_<NNN>_<description>` and in that function/method docstring. Module-level functions and class-based pytest methods are both supported. Multiple test functions may cover one Case ID; assertions come only from 预期结果/Expected Results and setup comes only from 前置条件 plus any optional 测试数据/自动化映射 or their legacy English aliases.",
+            "Name each generated pytest file so it corresponds one-to-one with its source Markdown module file: for each `testcase/md/<module>.md` (excluding README.md), emit exactly one `testcase/test_<module>.py`. The <module> stem is the Markdown filename without the `.md` extension, lowercased and with non-alphanumeric characters replaced by underscores. For example, `testcase/md/resource_notes.md` maps to `testcase/test_resource_notes.py`, and `testcase/md/order-api.md` maps to `testcase/test_order_api.py`. Never merge multiple Markdown modules into one pytest file, never split one module across several files, and never invent pytest filenames unrelated to the Markdown modules.",
             "Generate a reusable HTTP logging helper (or equivalent client wrapper) and call it for every interface request. The request log must include method, URL/path, and request parameters (query plus JSON/body/payload summary). The response log must include status code and response result (JSON/text/body summary), and both records must be visible in pytest stdout/stderr without changing assertions.",
             "Compare timestamps and other semantically equivalent protocol values by parsed meaning, not byte-for-byte serialization. In particular, normalize valid ISO-8601 instants before equality/order assertions so differences such as omitted trailing fractional seconds do not create TestBug failures; preserve exact-string assertions only when the Markdown explicitly requires representation equality.",
             "Before logging, recursively redact sensitive keys and header values including authorization, proxy-authorization, cookie, set-cookie, token, password, secret, api key and credentials. Never print full Authorization/Cookie values. Apply bounded truncation to serialized request and response bodies (with an explicit truncation marker) so large payloads cannot flood pytest or report artifacts.",
@@ -3981,6 +3998,7 @@ function buildMultiPerspectiveReviewNodes(input) {
                 commandSource: "inline",
                 fallbackCommands: [aggregateScript],
                 finalFullRequired: true,
+                commandTimeoutMs: 60_000,
             }),
             cwd: ".",
             timeoutMs: 60000,
@@ -4165,6 +4183,7 @@ function buildKnowledgeSyncValidateNode(sources, featureId) {
                 commandSource: "inline",
                 fallbackCommands: [validateScript],
                 finalFullRequired: true,
+                commandTimeoutMs: 120_000,
             }),
             cwd: ".",
             timeoutMs: 120000,
@@ -4385,6 +4404,7 @@ function buildKgBootstrapPreflightNode(sources) {
                 quota: "full",
                 commandSource: "inline",
                 fallbackCommands: [script],
+                commandTimeoutMs: 60_000,
             }),
             cwd: ".",
             timeoutMs: 60000,
@@ -4432,6 +4452,7 @@ function buildKgBootstrapInventoryNode(sources) {
                 quota: "full",
                 commandSource: "inline",
                 fallbackCommands: [script],
+                commandTimeoutMs: 120_000,
             }),
             cwd: ".",
             timeoutMs: 120000,
@@ -4526,6 +4547,7 @@ function buildKgBootstrapValidateNode(sources) {
                 commandSource: "inline",
                 fallbackCommands: [script],
                 finalFullRequired: true,
+                commandTimeoutMs: 120_000,
             }),
             cwd: ".",
             timeoutMs: 120000,
@@ -4653,6 +4675,7 @@ function buildKgBootstrapPromoteNode(sources) {
                 commandSource: "inline",
                 fallbackCommands: [script],
                 finalFullRequired: true,
+                commandTimeoutMs: 120_000,
             }),
             cwd: ".",
             timeoutMs: 120000,
@@ -4718,6 +4741,7 @@ function buildKgBootstrapMaterializeNode(sources) {
                 commandSource: "inline",
                 fallbackCommands: [script],
                 finalFullRequired: true,
+                commandTimeoutMs: 120_000,
             }),
             cwd: ".",
             timeoutMs: 120000,
@@ -4774,7 +4798,22 @@ export async function requireManagedTaskContractBinding(input) {
         taskId: input.taskId,
     });
     if (state.effectiveStatus !== "managed" || !state.ref) {
-        const err = new Error(`dag generate requires managed Task Contract for task ${input.taskId} (effectiveStatus=${state.effectiveStatus}); run: loop-agent task contract adopt|apply --task ${input.taskId} ...`);
+        const expectedRevision = state.ref?.revision ?? 0;
+        const observedCanonicalHash = state.observedCanonicalHash ?? "<observedCanonicalHash-from-show>";
+        const recovery = state.effectiveStatus === "transaction-incomplete"
+            ? `Recover the unfinished transaction first: loop-agent task contract recover --task ${input.taskId} --json`
+            : [
+                "After reviewing the show JSON, either adopt the current on-disk facts:",
+                `loop-agent task contract adopt --task ${input.taskId} --expected-revision ${expectedRevision} --expected-observed-hash ${observedCanonicalHash} --request-id <unique-request-id> --request-payload-sha256 <sha256-of-this-adopt-request> --json`,
+                "or apply a reviewed draft:",
+                `loop-agent task contract apply --task ${input.taskId} --input <reviewed-draft.json> --expected-revision ${expectedRevision} --expected-observed-hash ${observedCanonicalHash} --request-id <unique-request-id> --request-payload-sha256 <sha256-of-input-file> --json`,
+            ].join(" ");
+        const err = new Error([
+            `dag generate requires managed Task Contract for task ${input.taskId} (effectiveStatus=${state.effectiveStatus})`,
+            `Inspect current state: loop-agent task contract show ${input.taskId} --json`,
+            recovery,
+            "Do not auto-adopt or auto-apply before confirming the Task Contract facts and concurrency fields.",
+        ].join("; "));
         err.code =
             state.effectiveStatus === "externally-modified"
                 ? "EXTERNALLY_MODIFIED"
@@ -4828,10 +4867,6 @@ function synthesizeInMemoryTaskContractBinding(sources) {
     };
 }
 async function resolveTaskContractBindingForGenerate(sources) {
-    // Real CLI generation freezes the managed ref. Direct builder unit tests
-    // may provide repoRoot-shaped fixture paths without a persisted task; keep
-    // their deterministic synthetic binding path so topology/prompt tests remain
-    // isolated from contract storage.
     if (sources.repoRoot) {
         const state = await observeTaskContract({
             repoRoot: sources.repoRoot,
@@ -4986,7 +5021,12 @@ function buildReviewVerdictRecoveryNode(sources) {
     return {
         id: "review-verdict-recovery-pi",
         depends_on: ["review-pi"],
+        // AC3: recovery is a read-only failure-aware node. It runs after review-pi
+        // settles, consuming either normal FINISHED output or an explicitly tolerated
+        // ERROR (for example protocol-invalid after retries). It must not bypass the
+        // writer, write guard, or final shell gate, which remain fail-closed.
         role: "reviewer",
+        failureAwareDependsOn: ["review-pi"],
         executor: "pi",
         complexity: "LOW",
         writePolicy: "read-only",
@@ -5271,7 +5311,7 @@ function buildWriteSetGateNode(sources) {
 }
 function buildSoftVerifyNode(sources) {
     const implementId = implementationNodeId();
-    const strategy = resolveDagVerifyStrategy(sources.taskConfig);
+    const strategy = resolveDagVerifyStrategy(sources.taskConfig, "1");
     const fallbackCommands = ["npm run typecheck"];
     const commands = buildVerifyShellCommands({
         repoRoot: sources.repoRoot,
@@ -5297,9 +5337,10 @@ function buildSoftVerifyNode(sources) {
                 commandSource: sources.verifyCommands ? "adapter" : "inline",
                 commands: sources.verifyCommands?.intermediate,
                 fallbackCommands,
+                commandTimeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
             }),
             cwd: ".",
-            timeoutMs: 300000,
+            timeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
         },
     };
 }
@@ -5309,6 +5350,7 @@ function buildProcessSupervisorNode(sources) {
     return {
         id: "process-supervisor-pi",
         depends_on: ["soft-verify-shell", implementId],
+        failureAwareDependsOn: ["soft-verify-shell"],
         role: "supervisor",
         executor: "pi",
         complexity: "HIGH",
@@ -5415,9 +5457,10 @@ function buildHardVerifyNode(sources) {
                 commands: sources.verifyCommands?.final,
                 fallbackCommands,
                 finalFullRequired: true,
+                commandTimeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
             }),
             cwd: ".",
-            timeoutMs: 300000,
+            timeoutMs: DEFAULT_VERIFY_TIMEOUT_MS,
         },
     };
 }
@@ -5466,7 +5509,7 @@ function buildSupervisedHybridDag(standard, sources) {
         objective: `${standard.objective ?? ""}\n\nRoute: supervised implementation DAG selected by workflowPolicy/governanceProfile or explicit CLI profile.`.trim(),
         globalConstraints: supervisedConstraints,
         convergence: sources.taskConfig.convergence,
-        verifyStrategy: resolveDagVerifyStrategy(sources.taskConfig),
+        verifyStrategy: resolveDagVerifyStrategy(sources.taskConfig, "1"),
         tasks: [
             cloneTask(contract),
             cloneTask(scoutSrc),
@@ -5536,6 +5579,12 @@ export async function writeHybridDagDraft(sources, outputPath, options = {}) {
         ? await prepareFrontendMockSources(sources)
         : sources;
     const spec = await buildHybridDagForTemplate(preparedSources, template);
+    if (preparedSources.repoRoot && dagHasWriterExecution(spec)) {
+        await requireManagedTaskContractBinding({
+            repoRoot: preparedSources.repoRoot,
+            taskId: preparedSources.taskId,
+        });
+    }
     await writeFile(outputPath, `${JSON.stringify(spec, null, 2)}\n`, "utf-8");
     return {
         taskId: sources.taskId,