@lazyingart/agintiflow 0.20.302 → 0.20.304

package/docs/supervision-campaign-ledger.md

@@ -650,3 +650,28 @@ append-only repairs when replacement is required, and caps repair schemas so a
 small local model receives only the actionable tool surface. Focused coding,
 truthful-completion, writing-routing, progressive-tool, document-quality, and
 dynamic-budget suites pass. These fixes are prepared for AgInTiFlow `0.20.302`.
+
+### Host-scoped deliverables without false source mutation
+
+LabCanvas research run `browser-agent-stale-ui-research-085` showed that
+DeepSeek had already completed current web research and written valid task
+artifacts, but AgInTiFlow rejected `finish` because every file-producing task
+inherited a fresh project-source mutation requirement. The artifact directory
+was intentionally excluded from project mutation accounting, so the runtime
+could never satisfy its own gate and eventually reached the step limit.
+
+The runtime now distinguishes a requested deliverable inside the host-owned
+artifact root from a requested source-code change. Reports, notes, PDFs,
+figures, CAD, presentations, and similar scoped outputs may complete from their
+verified artifact evidence without fabricating a repository edit. A request to
+fix, implement, or modify project source still requires a fresh project
+mutation, even when it also asks for a sidecar report.
+
+Fresh LabCanvas retest `browser-agent-stale-ui-research-086` used the same
+imperfect prompt and a new persistent AgInTi session. It recovered a malformed
+DeepSeek tool call, ran web research, wrote and registered two substantial
+Markdown artifacts with current primary-source citations, separated
+demonstrated evidence from inferred recommendations, and finished normally.
+The retained trace contains no completion-evidence rejection or forced
+source-mutation loop. Focused completion, evidence, progressive-tool, runtime,
+and dynamic-budget suites plus the full npm suite pass for `0.20.303`.

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@lazyingart/agintiflow",
-  "version": "0.20.302",
+  "version": "0.20.304",
   "type": "module",
   "description": "AgInTiFlow is a project-aware agent workspace for hybrid wet-dry R&D, hardware-aware intelligence, software automation, and industrial workflows.",
   "license": "Apache-2.0",
@@ -230,7 +230,7 @@
     "publish:env": "node scripts/npm-publish-from-env.js publish --access public",
     "publish:env:whoami": "node scripts/npm-publish-from-env.js whoami",
     "pretest": "npm run smoke:planning-timeout-recovery",
-    "test": "npm run check && npm run smoke:integration-production-mount && npm run smoke:integration-production-runtime-bundle && npm run smoke:integration-retained-idempotency-store && npm run smoke:integration-storage-authority && npm run smoke:integration-retained-durable-common && npm run smoke:integration-retained-file-lock && npm run smoke:integration-retained-event-ledger && npm run smoke:integration-retained-event-ledger-bundle && npm run smoke:integration-retained-repository-kernel && npm run smoke:integration-retained-session-state-store && npm run smoke:integration-retained-native-session-repository-state && npm run smoke:integration-retained-native-execution-evidence && npm run smoke:integration-retained-runtime-repository-surface && npm run smoke:integration-retained-runtime-repository-phase-b && npm run smoke:integration-runtime-repository-contract && npm run smoke:integration-runtime-authority && npm run smoke:localllm-provider && npm run smoke:localllm-model-tiers && npm run smoke:localllm-code-routing && npm run smoke:local-failure-recovery && npm run smoke:localllm-auto-max && npm run smoke:local-resource-policy && npm run smoke:context-budget-recovery && npm run smoke:session-runtime && npm run smoke:runtime-core && npm run smoke:progressive-tools && npm run smoke:truthful-completion && npm run smoke:document-artifact-quality && npm run smoke:writing-specialist-routing && npm run smoke:workspace-secret-false-positive && npm run smoke:provider-handoff && npm run eval:local-first-agent && npm run eval:provider-attribution && npm run smoke:runtime-compat && npm run smoke:autoupdate && npm run smoke:web-api && npm run smoke:math-rendering && npm run smoke:web-ui && npm run smoke:web-autostart && npm run smoke:webapp-command && npm run smoke:web-port-fallback && npm run smoke:docker-command && npm run smoke:coding-tools && npm run smoke:dynamic-step-budget && npm run smoke:execution-policy && npm run smoke:aaps-adapter && npm run smoke:auxiliary-tools && npm run smoke:perception-research && npm run smoke:deep-research && npm run smoke:public-research && npm run smoke:safe-chat && npm run smoke:auth && npm run smoke:agentlink && npm run smoke:canvas-artifacts && npm run smoke:capabilities && npm run smoke:mcp && npm run smoke:model-roles && npm run smoke:platform && npm run smoke:permission-modes && npm run smoke:skills && npm run smoke:skillmesh && npm run smoke:supervision-ledger && npm run smoke:tmux-tools && npm run smoke:long-jobs && npm run smoke:run-stdin && npm run smoke:cli-chat && npm run smoke:inbox",
+    "test": "npm run check && npm run smoke:integration-production-mount && npm run smoke:integration-production-runtime-bundle && npm run smoke:integration-retained-idempotency-store && npm run smoke:integration-storage-authority && npm run smoke:integration-retained-durable-common && npm run smoke:integration-retained-file-lock && npm run smoke:integration-retained-event-ledger && npm run smoke:integration-retained-event-ledger-bundle && npm run smoke:integration-retained-repository-kernel && npm run smoke:integration-retained-session-state-store && npm run smoke:integration-retained-native-session-repository-state && npm run smoke:integration-retained-native-execution-evidence && npm run smoke:integration-retained-runtime-repository-surface && npm run smoke:integration-retained-runtime-repository-phase-b && npm run smoke:integration-runtime-repository-contract && npm run smoke:integration-runtime-authority && npm run smoke:localllm-provider && npm run smoke:localllm-model-tiers && npm run smoke:localllm-code-routing && npm run smoke:local-failure-recovery && npm run smoke:localllm-auto-max && npm run smoke:local-resource-policy && npm run smoke:context-budget-recovery && npm run smoke:session-runtime && npm run smoke:runtime-core && npm run smoke:progressive-tools && npm run smoke:scoped-artifact-research && npm run smoke:truthful-completion && npm run smoke:document-artifact-quality && npm run smoke:writing-specialist-routing && npm run smoke:workspace-secret-false-positive && npm run smoke:provider-handoff && npm run eval:local-first-agent && npm run eval:provider-attribution && npm run smoke:runtime-compat && npm run smoke:autoupdate && npm run smoke:web-api && npm run smoke:math-rendering && npm run smoke:web-ui && npm run smoke:web-autostart && npm run smoke:webapp-command && npm run smoke:web-port-fallback && npm run smoke:docker-command && npm run smoke:coding-tools && npm run smoke:dynamic-step-budget && npm run smoke:execution-policy && npm run smoke:aaps-adapter && npm run smoke:auxiliary-tools && npm run smoke:perception-research && npm run smoke:deep-research && npm run smoke:public-research && npm run smoke:safe-chat && npm run smoke:auth && npm run smoke:agentlink && npm run smoke:canvas-artifacts && npm run smoke:capabilities && npm run smoke:mcp && npm run smoke:model-roles && npm run smoke:platform && npm run smoke:permission-modes && npm run smoke:skills && npm run smoke:skillmesh && npm run smoke:supervision-ledger && npm run smoke:tmux-tools && npm run smoke:long-jobs && npm run smoke:run-stdin && npm run smoke:cli-chat && npm run smoke:inbox",
     "pack:dry-run": "npm pack --dry-run",
     "smoke:capabilities": "node scripts/smoke-capabilities.js"
   },

package/scripts/smoke-scs-evidence-visibility.js

@@ -205,6 +205,20 @@ const readOnlyReviewContract = deriveScsTaskContract({
   taskProfile: "review",
 });
 
+const readOnlyOperatorAuditRequest =
+  "This is a short imperfect operator audit. Check these three things without changing source files or sending messages: 1) run curl -fsS http://127.0.0.1:8008/readyz and state whether LocalLLM is ready; 2) read configs/model-policy.json and name the configured AgInTi provider order; 3) run git status --short and say only whether tracked source changes exist, without treating ignored runtime output as source. Cover all three numbered items in one concise answer and write the normal task result.";
+const readOnlyOperatorAuditContract = deriveScsTaskContract({
+  goal: [
+    readOnlyOperatorAuditRequest,
+    `AGINTI_EVIDENCE_SCOPE_JSON: ${JSON.stringify({
+      mode: "task",
+      request: readOnlyOperatorAuditRequest,
+      artifact_root: "/tmp/aginti-read-only-audit",
+    })}`,
+  ].join("\n\n"),
+  taskProfile: "auto",
+});
+
 const explicitReviewRepairContract = deriveScsTaskContract({
   goal: [
     "Review focus: inspect changed files and fix confirmed defects.",
@@ -374,6 +388,34 @@ assert.equal(
   true,
   "an explicit review-and-fix focus lost its positive mutation request"
 );
+assert.equal(
+  readOnlyOperatorAuditContract.requiresWorkspaceMutation,
+  false,
+  "a host-managed task response was mistaken for a workspace mutation"
+);
+assert.equal(
+  readOnlyOperatorAuditContract.requiresFileMutation,
+  false,
+  "an observational source-status mention was mistaken for a source mutation"
+);
+assert(
+  readOnlyOperatorAuditContract.requiredEvidence.some(
+    (item) => item.category === "command"
+  ) &&
+    !readOnlyOperatorAuditContract.requiredEvidence.some(
+      (item) => ["artifact", "file"].includes(item.category)
+    ),
+  "the phrase 'cover all items' invented cover-art or file evidence"
+);
+assert(
+  readOnlyOperatorAuditContract.forbiddenActions.some((item) =>
+    /changing source files or sending messages/i.test(item)
+  ) &&
+    !readOnlyOperatorAuditContract.forbiddenActions.some((item) =>
+      /run curl/i.test(item)
+    ),
+  "a colon after a without-clause swallowed the following command"
+);
 assert(
   pageSafeReportContract.requiredEvidence.some((item) => item.category === "file"),
   "a scoped existing-report edit did not require file evidence"

package/scripts/smoke-web-ui.js

@@ -138,7 +138,9 @@ try {
   if ((await page.locator('[data-workspace-file="notes/workspace-ui.md"]').getAttribute("draggable")) !== "true") {
     throw new Error("workspace file rows are not draggable");
   }
-  await page.locator("#workspace-automap").click();
+  const automapButton = page.locator("#workspace-automap");
+  await automapButton.scrollIntoViewIfNeeded();
+  await automapButton.click();
   await page.locator('[data-workspace-lane="docs"]').click();
   await page.waitForSelector('[data-workspace-file="notes/workspace-ui.md"]', { timeout: 10000 });
   await page.locator('[data-workspace-file="notes/workspace-ui.md"]').click();

package/src/agent-runner.js

@@ -4289,6 +4289,8 @@ export function resetSameTaskExecutionContract(state = {}, revision = 0) {
     startedMutationRevision,
     requiresWorkspaceMutation: currentTurnContract.requiresWorkspaceMutation === true,
     requiresFileMutation: currentTurnContract.requiresFileMutation === true,
+    scopedArtifactDeliverable:
+      currentTurnContract.scopedArtifactDeliverable === true,
     requiresSourceGrounding: Boolean(
       currentTurnContract.requiresSourceGrounding === true ||
         (
@@ -14666,8 +14668,12 @@ export function completionTaskContract(config = {}, state = {}) {
           0
       )
     );
+    const scopedArtifactDeliverable = currentContract
+      ? currentContract.scopedArtifactDeliverable === true
+      : activeExecutionContract.scopedArtifactDeliverable === true;
     const requiresFreshMutation = Boolean(
-      currentContract?.requiresFileMutation || activeExecutionContract.requiresFileMutation
+      (currentContract?.requiresFileMutation || activeExecutionContract.requiresFileMutation) &&
+        !scopedArtifactDeliverable
     );
     const minimumMutationRevision = requiresFreshMutation
       ? startedMutationRevision + 1
@@ -14720,6 +14726,7 @@ export function completionTaskContract(config = {}, state = {}) {
       requiresFileMutation: authoritativeReadOnlyContinuation
         ? false
         : Boolean(contract.requiresFileMutation || currentContract?.requiresFileMutation),
+      scopedArtifactDeliverable,
       requiresSourceGrounding: authoritativeReadOnlyContinuation
         ? false
         : Boolean(contract.requiresSourceGrounding || currentContract?.requiresSourceGrounding),

package/src/scs-evidence.js

@@ -1275,8 +1275,19 @@ function goalRequestsExplicitTestMutation(text = "") {
   );
 }
 
+function stripHostManagedResponseNarration(goal = "") {
+  return String(goal || "").replace(
+    /\b(?:and\s+)?(?:write|return|provide)\s+(?:(?:a|the)\s+)?(?:(?:concise|final|normal|structured)\s+)?(?:(?:agent|task)\s+)?(?:answer|response|result)\b(?!\s+(?:as|at|file|in|into|json|markdown|pdf|text|to|under)\b)/gi,
+    ""
+  );
+}
+
 function goalRequestsWorkspaceMutation(goal = "", taskProfile = "") {
-  const text = normalizedText(stripCompletedWorkNarration(stripForbiddenLanguage(goal)));
+  const text = normalizedText(
+    stripHostManagedResponseNarration(
+      stripCompletedWorkNarration(stripForbiddenLanguage(goal))
+    )
+  );
   const explicitAddMutation =
     goalRequestsExplicitTestMutation(text) ||
     /\badd\b(?:\s+(?:a|an|the|new|additional|specific))*\s+(?:code|documents?|files?|notes?|readme|scripts?|source|workspace)\b/.test(
@@ -1316,6 +1327,20 @@ function goalRequestsFileMutation(goal = "", taskProfile = "") {
   );
 }
 
+function goalRequestsScopedArtifactDeliverable(goal = "") {
+  const text = normalizedText(stripCompletedWorkNarration(stripForbiddenLanguage(goal)));
+  const deliverable =
+    /\b(?:build|compile|create|draft|export|generate|make|prepare|produce|render|save|write)\b[^.\n;]{0,180}\b(?:artifact|brief|cad|diagram|document|figure|image|markdown|memo|model|note|paper|pcb|pdf|presentation|prompt|report|research note|slide deck|slides?|spreadsheet|story|summary|transcript|video)\b/.test(text) ||
+    /\b(?:artifact|brief|cad|diagram|document|figure|image|markdown|memo|model|note|paper|pcb|pdf|presentation|prompt|report|research note|slide deck|slides?|spreadsheet|story|summary|transcript|video)\b[^.\n;]{0,120}\b(?:build|compile|create|draft|export|generate|make|prepare|produce|render|save|write)\b/.test(text) ||
+    /(?:创建|生成|保存|编写|撰写|制作|导出|准备)[^。；\n]{0,100}(?:报告|笔记|摘要|文档|论文|图表|图片|模型|演示文稿|幻灯片|表格|故事|提示词|转录|视频)/.test(text);
+  if (!deliverable) return false;
+  const projectSourceMutation =
+    /\b(?:change|debug|edit|fix|implement|improve|modify|patch|refactor|repair|replace|rewrite|update)\b[^.\n;]{0,140}\b(?:app|application|code|codebase|implementation|library|package|repo|repository|runtime|source|src\/)\b/.test(text) ||
+    /\b(?:app|application|code|codebase|implementation|library|package|repo|repository|runtime|source|src\/)\b[^.\n;]{0,140}\b(?:change|debug|edit|fix|implement|improve|modify|patch|refactor|repair|replace|rewrite|update)\b/.test(text) ||
+    /(?:修改|修复|实现|改进|重构|调试|替换|更新)[^。；\n]{0,100}(?:代码|代码库|实现|运行时|应用|仓库|源码)/.test(text);
+  return !projectSourceMutation;
+}
+
 function goalRequestsTestExecution(goal = "") {
   const text = normalizedText(stripCompletedWorkNarration(stripForbiddenLanguage(goal)));
   return (
@@ -1434,7 +1459,12 @@ function inferRequirementCategories(goal = "", taskProfile = "", acceptanceCrite
   if (goalRequestsTestExecution(positiveGoal)) {
     categories.add("test");
   }
-  if (textHas(mandatoryEvidenceText, /\b(artifact|canvas|pdf|image|video|screenshot|cover|plot|chart|figure|docx|archive|copy to|export|generated|generate|draft)\b/) || /输出|产物|图片|视频|截图|封面|生成/.test(mandatoryEvidenceText)) {
+  if (
+    textHas(
+      mandatoryEvidenceText,
+      /\b(artifact|canvas|pdf|image|video|screenshot|plot|chart|figure|docx|archive|copy to|export|generated|generate|draft)\b|\b(?:album|book|paper|report|video)\s+cover\b|\bcover\s+(?:art|design|image)\b/
+    ) || /输出|产物|图片|视频|截图|封面|生成/.test(mandatoryEvidenceText)
+  ) {
     categories.add("artifact");
   }
   if (
@@ -1499,14 +1529,14 @@ function inferForbiddenActions(goal = "") {
   const text = String(goal || "");
   const forbidden = [];
   const isAction = (value = "") =>
-    /\b(use|open|click|browse|browser|upload|attach|submit|publish|deploy|run|execute|install|delete|remove|commit|push|call|api|alter|change|edit|fix|modify|patch|repair|rewrite|touch|write)\b/i.test(
+    /\b(use|open|click|browse|browser|upload|attach|submit|publish|deploy|run|execute|install|delete|remove|commit|push|call|api|alter|chang(?:e|ing)|edit(?:ing)?|fix(?:ing)?|modif(?:y|ying)|patch(?:ing)?|repair(?:ing)?|rewrit(?:e|ing)|send(?:ing)?|touch(?:ing)?|writ(?:e|ing))\b/i.test(
       value
     ) || /浏览器|网页|打开|点击|上传|提交|发布|部署|运行|执行|安装|删除|复制|移动|修改|编辑|修复|改写|写入|提交代码|推送|调用|API/.test(value);
   const patterns = [
     { re: /\b(do not|don't|dont|never|no need to)\s+([^.\n;]+)/gi, prefix: "User forbids" },
     // A sentence such as "without changing X, run tests and commit" starts a
     // positive instruction after the comma. Keep only the local without-clause.
-    { re: /\bwithout\s+([^.,\n;]+)/gi, prefix: "User forbids" },
+    { re: /\bwithout\s+([^.,:\uFF1A\n;]+)/gi, prefix: "User forbids" },
     { re: /不要([^。\n；]+)/g, prefix: "User forbids" },
     { re: /禁止([^。\n；]+)/g, prefix: "User forbids" },
   ];
@@ -2029,6 +2059,9 @@ export function deriveScsTaskContract({ goal = "", taskProfile = "", acceptanceC
     positiveEvidenceGoal,
     taskProfile
   );
+  const scopedArtifactDeliverable = Boolean(
+    artifactRoot && goalRequestsScopedArtifactDeliverable(evidenceGoal)
+  );
   return {
     version: 1,
     outcome: compact(evidenceGoal || "Complete the requested task.", 500),
@@ -2038,6 +2071,7 @@ export function deriveScsTaskContract({ goal = "", taskProfile = "", acceptanceC
     forbiddenActions: inferForbiddenActions(evidenceGoal),
     exactOutputPaths,
     requiredArtifactKinds,
+    scopedArtifactDeliverable,
     excludedOutputPaths,
     artifactRoot,
     exactInputPaths,