@tea-agent/loop-agent 0.33.6-beta.0 → 0.33.6

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 (64) hide show
  1. package/CHANGELOG.md +29 -4
  2. package/dist/application/task-lifecycle/advance.js +254 -4
  3. package/dist/application/task-lifecycle/gates.js +50 -0
  4. package/dist/application/task-lifecycle/observe.js +11 -2
  5. package/dist/commands/init-upgrade.js +32 -1
  6. package/dist/commands/init.js +94 -3
  7. package/dist/executors/shell-write-guard.js +26 -8
  8. package/dist/shared/operator/capabilities.js +72 -42
  9. package/dist/shared/resilient-git.js +133 -0
  10. package/dist/task/source-prepare/artifact-meta.js +137 -0
  11. package/dist/task/source-prepare/index.js +2 -0
  12. package/dist/task/source-prepare/parse-intent.js +58 -10
  13. package/dist/task/source-prepare/prepare.js +180 -16
  14. package/dist/task/source-prepare/reference-integrity.js +18 -2
  15. package/dist/task/source-prepare/semantic-intake.js +404 -0
  16. package/dist/worker/console/app-data.js +2 -0
  17. package/dist/worker/console/chat/chat-event-store.js +190 -25
  18. package/dist/worker/console/chat/pi-console-config.js +250 -32
  19. package/dist/worker/console/chat/pi-runtime.js +625 -71
  20. package/dist/worker/console/chat/resource-loader.js +5 -4
  21. package/dist/worker/console/chat/routes.js +324 -146
  22. package/dist/worker/console/chat/runtime-context.js +48 -12
  23. package/dist/worker/console/chat/runtime-selection.js +59 -0
  24. package/dist/worker/console/chat/shortcuts.js +1 -0
  25. package/dist/worker/console/chat/tool-adapter.js +9 -3
  26. package/dist/worker/console/chat/tools.js +5 -1
  27. package/dist/worker/console/dag-execution-receipt.js +380 -0
  28. package/dist/worker/console/operator-actions.js +559 -68
  29. package/dist/worker/console/server.js +8 -15
  30. package/dist/worker/console/static/assets/index-BUOLppPr.js +28 -0
  31. package/dist/worker/console/static/assets/index-C1KzazY5.css +1 -0
  32. package/dist/worker/console/static/index.html +2 -2
  33. package/dist/worker/console/static-src/operator-chat/chat-sse-events.js +45 -8
  34. package/dist/worker/console/static-src/operator-chat/refs.js +9 -0
  35. package/dist/worker/console/static-src/operator-chat/runtime-snapshot-store.js +257 -0
  36. package/dist/worker/console/static-src/operator-chat/useChatSessions.js +16 -0
  37. package/dist/worker/console/static-src/operator-chat/useChatStream.js +210 -184
  38. package/dist/worker/console/static-src/operator-chat/useChatThread.js +49 -5
  39. package/dist/worker/console/static-src/operator-chat/useComposer.js +17 -0
  40. package/dist/worker/console/static-src/operator-chat/useRuntimeControls.js +225 -74
  41. package/dist/worker/console/static-src/operator-chat/useRuntimeSnapshot.js +196 -0
  42. package/dist/worker/delivery/final-verification.js +13 -5
  43. package/dist/worker/delivery/package.js +31 -19
  44. package/dist/worker/delivery/verification-bundle.js +6 -4
  45. package/dist/worker/observe/static/operator-chrome.css +5 -2
  46. package/dist/worker/observe/static/operator-chrome.js +6 -1
  47. package/dist/worker/observe/static/styles.css +39 -9
  48. package/dist/workflows/dag/backend-test-case-coverage-analysis.js +2 -27
  49. package/dist/workflows/dag/backend-test-module-stem.js +0 -5
  50. package/dist/workflows/dag/backend-test-writer-completeness.js +16 -47
  51. package/dist/workflows/dag/dynamic-runtime/map.js +8 -24
  52. package/dist/workflows/dag/frontend-worktree-diff.js +12 -27
  53. package/dist/workflows/dag/init-hybrid.js +12 -20
  54. package/dist/workflows/dag/types.js +0 -7
  55. package/dist/workflows/dag/workspace-checkpoint.js +8 -27
  56. package/docs/templates/backend-test-dag.json +10 -10
  57. package/harness.json +1 -1
  58. package/package.json +1 -1
  59. package/skills/loop-agent/references/command-reference.md +3 -1
  60. package/skills/loop-agent/references/source-and-plan-practice.md +13 -0
  61. package/skills/loop-agent/references/task-workflow.md +4 -0
  62. package/dist/worker/console/chat/instruction-skills.js +0 -217
  63. package/dist/worker/console/static/assets/index-CnUXAqxG.css +0 -1
  64. package/dist/worker/console/static/assets/index-CteJFFL2.js +0 -29
@@ -1,8 +1,7 @@
1
1
  import { createHash, randomUUID } from "node:crypto";
2
- import { execFile } from "node:child_process";
3
2
  import { mkdir, readFile, realpath, rename, rm, writeFile } from "node:fs/promises";
4
3
  import path from "node:path";
5
- import { promisify } from "node:util";
4
+ import { isResilientGitCommandError, runResilientGitCommand } from "../../shared/resilient-git.js";
6
5
  import YAML from "yaml";
7
6
  import { z } from "zod";
8
7
  import { controllerIdentitiesMatch, LOOP_AGENT_PACKAGE_NAME, } from "../loop-agent/loop-agent-client.js";
@@ -13,7 +12,6 @@ import { validateFeatureTaskGraph } from "../task-graph/validate.js";
13
12
  import { taskSpecSchema } from "../task-spec/schema.js";
14
13
  import { gitTransactionRecordSchema, transactionRecordPath } from "./git-transaction.js";
15
14
  import { readVerifiedBundle, verifyBundleCurrentStateBindings, verifyBundleOutcomes, verifyBundleTaskSpecBindings } from "./verification-bundle.js";
16
- const execFileAsync = promisify(execFile);
17
15
  const hashedRefSchema = z.object({ path: z.string().min(1), sha256: z.string().regex(/^[a-f0-9]{64}$/) }).strict();
18
16
  export const acceptanceCoverageArtifactSchema = z.object({
19
17
  schemaVersion: z.literal(1),
@@ -91,9 +89,9 @@ export async function prepareFeatureDelivery(input) {
91
89
  const now = input.now ?? new Date();
92
90
  const record = await readTransaction(repoRoot, featureId, blockers);
93
91
  if (record) {
94
- const branch = await git(repoRoot, ["branch", "--show-current"]).catch(() => "");
95
- const head = await git(repoRoot, ["rev-parse", "HEAD"]).catch(() => "");
96
- const status = await gitRaw(repoRoot, ["status", "--porcelain=v1", "--untracked-files=all"]).catch(() => "git-error");
92
+ const branch = await git(repoRoot, ["branch", "--show-current"]).catch(rethrowInfrastructureOr(""));
93
+ const head = await git(repoRoot, ["rev-parse", "HEAD"]).catch(rethrowInfrastructureOr(""));
94
+ const status = await gitRaw(repoRoot, ["status", "--porcelain=v1", "--untracked-files=all"]).catch(rethrowInfrastructureOr("git-error"));
97
95
  if (branch !== record.branch)
98
96
  blockers.push(`current branch must be ${record.branch}`);
99
97
  if (head !== record.lastCheckpoint)
@@ -202,11 +200,11 @@ export async function validateDeliveryPackage(input) {
202
200
  blockers.push("Delivery Feature ownership mismatch");
203
201
  if (manifest.branch !== record.branch || manifest.baseBranch !== record.baseBranch || manifest.baseSha !== record.baseCommit || manifest.headSha !== record.lastCheckpoint)
204
202
  blockers.push("Delivery manifest does not match the Git transaction");
205
- const branch = await git(repoRoot, ["branch", "--show-current"]).catch(() => "");
206
- const head = await git(repoRoot, ["rev-parse", "HEAD"]).catch(() => "");
203
+ const branch = await git(repoRoot, ["branch", "--show-current"]).catch(rethrowInfrastructureOr(""));
204
+ const head = await git(repoRoot, ["rev-parse", "HEAD"]).catch(rethrowInfrastructureOr(""));
207
205
  if (branch !== record.branch || head !== record.lastCheckpoint)
208
206
  blockers.push("current Git position does not match Delivery");
209
- const rawCloseoutStatus = await gitRaw(repoRoot, ["status", "--porcelain=v1", "--untracked-files=all"]).catch(() => "git-error");
207
+ const rawCloseoutStatus = await gitRaw(repoRoot, ["status", "--porcelain=v1", "--untracked-files=all"]).catch(rethrowInfrastructureOr("git-error"));
210
208
  const allowedDirtyPaths = new Set((input.allowedDirtyPaths ?? []).map((item) => item.replace(/\\/g, "/")));
211
209
  const closeoutStatus = rawCloseoutStatus.split("\n").filter(Boolean).filter((line) => !allowedDirtyPaths.has(line.slice(3).replace(/\\/g, "/"))).join("\n").trim();
212
210
  if (closeoutStatus)
@@ -495,23 +493,37 @@ async function writeDeliveryArtifacts(artifacts, coverage, manifest) {
495
493
  }
496
494
  }
497
495
  function renderPr(manifest) { return `# Summary\n\nFeature ${manifest.featureId} local Delivery Package.\n\n## Feature/base/head\n\n- Branch: ${manifest.branch}\n- Base: ${manifest.baseBranch} @ ${manifest.baseSha}\n- Head: ${manifest.headSha}\n\n## Completed Tasks\n\n${manifest.tasks.map((task) => `- ${task.taskId}: ${task.commit} (${task.changedFiles.join(", ")})`).join("\n")}\n\n## Acceptance Coverage\n\n- Required: ${manifest.acceptanceCoverage.requiredTotal}\n- Covered: ${manifest.acceptanceCoverage.covered}\n- Waived: ${manifest.acceptanceCoverage.waived}\n- Blocked: ${manifest.acceptanceCoverage.blocked}\n\n## Verification\n\n- QA: ${manifest.qa.evidence.path}\n- Final: ${manifest.finalVerification.path}\n\n## Changed Files\n\n${[...new Set(manifest.tasks.flatMap((task) => task.changedFiles))].map((file) => `- ${file}`).join("\n")}\n\n## Risks\n\n- high=${manifest.riskSummary.high}, medium=${manifest.riskSummary.medium}, low=${manifest.riskSummary.low}\n\n## Human Review Checklist\n\n- [ ] Review diff and commits\n- [ ] Confirm required AC coverage and waivers\n- [ ] Confirm QA and final verification evidence\n- [ ] Review risks\n- [ ] Decide whether to push or create a remote PR manually\n`; }
498
- async function git(repoRoot, args) { const result = await execFileAsync("git", args, { cwd: repoRoot, encoding: "utf8" }); return result.stdout.trim(); }
499
- async function gitRaw(repoRoot, args) { const result = await execFileAsync("git", args, { cwd: repoRoot, encoding: "utf8" }); return result.stdout; }
500
- async function gitOk(repoRoot, args) { try {
501
- await git(repoRoot, args);
502
- return true;
496
+ async function git(repoRoot, args) {
497
+ const result = await runResilientGitCommand({ cwd: repoRoot, args, readOnly: true });
498
+ if (result.code !== 0)
499
+ throw new Error(`git ${args.join(" ")} failed: ${result.stderr.trim() || result.stdout.trim() || `exit ${result.code}`}`);
500
+ return result.stdout.trim();
501
+ }
502
+ async function gitRaw(repoRoot, args) {
503
+ const result = await runResilientGitCommand({ cwd: repoRoot, args, readOnly: true });
504
+ if (result.code !== 0)
505
+ throw new Error(`git ${args.join(" ")} failed: ${result.stderr.trim() || result.stdout.trim() || `exit ${result.code}`}`);
506
+ return result.stdout;
507
+ }
508
+ async function gitOk(repoRoot, args) {
509
+ const result = await runResilientGitCommand({ cwd: repoRoot, args, readOnly: true });
510
+ return result.code === 0;
511
+ }
512
+ function rethrowInfrastructureOr(fallback) {
513
+ return (error) => {
514
+ if (isResilientGitCommandError(error))
515
+ throw error;
516
+ return fallback;
517
+ };
503
518
  }
504
- catch {
505
- return false;
506
- } }
507
519
  async function checkpointMatches(repoRoot, head, checkpoint, spec) {
508
520
  if (!await gitOk(repoRoot, ["merge-base", "--is-ancestor", checkpoint.commit, head]))
509
521
  return false;
510
- const messageText = await gitRaw(repoRoot, ["show", "-s", "--format=%B", checkpoint.commit]).catch(() => "");
522
+ const messageText = await gitRaw(repoRoot, ["show", "-s", "--format=%B", checkpoint.commit]).catch(rethrowInfrastructureOr(""));
511
523
  for (const trailer of [`Feature: ${spec.feature_id}`, `Task: ${spec.id}`, `Acceptance: ${spec.acceptance_refs.join(", ")}`, `Worker-Run: ${checkpoint.workerRunId}`])
512
524
  if (!messageText.split(/\r?\n/).includes(trailer))
513
525
  return false;
514
- const actual = (await gitRaw(repoRoot, ["diff-tree", "--no-commit-id", "--name-only", "-r", checkpoint.commit]).catch(() => "")).split(/\r?\n/).filter(Boolean).sort();
526
+ const actual = (await gitRaw(repoRoot, ["diff-tree", "--no-commit-id", "--name-only", "-r", checkpoint.commit]).catch(rethrowInfrastructureOr(""))).split(/\r?\n/).filter(Boolean).sort();
515
527
  return JSON.stringify(actual) === JSON.stringify([...checkpoint.changedFiles].sort());
516
528
  }
517
529
  async function hashedRefValid(repoRoot, ref) { try {
@@ -7,6 +7,7 @@ import { artifactMatchesEvidenceToken, isShellVerificationToken, } from "../outc
7
7
  import { readVerifiedOutcome } from "../outcomes/store.js";
8
8
  import { getTaskPoolRoot, readFeatureTaskPoolStates } from "../pool/run-store.js";
9
9
  import { workflowSchema } from "../task-spec/workflow-routing.js";
10
+ import { runResilientGitCommand } from "../../shared/resilient-git.js";
10
11
  import { gitTransactionRecordSchema, transactionRecordPath } from "./git-transaction.js";
11
12
  /**
12
13
  * Feature Verification Bundle v1 — a hash-bound aggregate of completed typed
@@ -533,10 +534,11 @@ async function readGitRecord(repoRoot, featureId) {
533
534
  }
534
535
  }
535
536
  async function defaultGit(repoRoot, args) {
536
- const { execFile } = await import("node:child_process");
537
- const { promisify } = await import("node:util");
538
- const execFileAsync = promisify(execFile);
539
- return (await execFileAsync("git", args, { cwd: repoRoot, encoding: "utf8" })).stdout.trim();
537
+ const result = await runResilientGitCommand({ cwd: repoRoot, args, readOnly: true });
538
+ if (result.code !== 0) {
539
+ throw new Error(`git ${args.join(" ")} failed: ${result.stderr.trim() || result.stdout.trim() || `exit ${result.code}`}`);
540
+ }
541
+ return result.stdout.trim();
540
542
  }
541
543
  // Re-export the workflow schema so consumers can validate typed workflows.
542
544
  export { workflowSchema };
@@ -71,8 +71,11 @@
71
71
  text-decoration: none;
72
72
  white-space: nowrap;
73
73
  }
74
- .och-brand svg {
75
- color: var(--och-orange);
74
+ .och-brand-mark {
75
+ display: block;
76
+ width: 32px;
77
+ height: 32px;
78
+ object-fit: contain;
76
79
  flex-shrink: 0;
77
80
  }
78
81
  .och-brand:hover {
@@ -353,7 +353,12 @@ export function mountOperatorChrome(root, config = {}) {
353
353
 
354
354
  const brand = el("a", "och-brand");
355
355
  brand.href = model.homeHref;
356
- icon(brand, "radar");
356
+ brand.setAttribute("aria-label", `${model.brand} 首页`);
357
+ const brandMark = el("img", "och-brand-mark");
358
+ brandMark.src = "/inspect/favicon.svg";
359
+ brandMark.alt = "";
360
+ brandMark.setAttribute("aria-hidden", "true");
361
+ brand.appendChild(brandMark);
357
362
  brand.appendChild(el("span", null, model.brand));
358
363
  global.appendChild(brand);
359
364
 
@@ -1617,7 +1617,11 @@ body.is-resizing-dag-graph {
1617
1617
  /* 历史失败已恢复/已覆盖:独立中性样式(不占用当前终态失败红色,不洗白事实)。 */
1618
1618
  .dag-trajectory-occurrence.is-recovered {
1619
1619
  border-color: color-mix(in srgb, var(--amber, #b58900) 55%, var(--hairline));
1620
- background: color-mix(in srgb, var(--amber-soft, #fff7d6) 55%, var(--surface));
1620
+ background: color-mix(
1621
+ in srgb,
1622
+ var(--amber-soft, #fff7d6) 55%,
1623
+ var(--surface)
1624
+ );
1621
1625
  }
1622
1626
  .dag-trajectory-recovered {
1623
1627
  font-size: 11px;
@@ -1647,9 +1651,14 @@ body.is-resizing-dag-graph {
1647
1651
  }
1648
1652
  .dag-trajectory-banner {
1649
1653
  padding: 8px 12px;
1650
- border: 1px solid color-mix(in srgb, var(--amber, #b58900) 50%, var(--hairline));
1654
+ border: 1px solid
1655
+ color-mix(in srgb, var(--amber, #b58900) 50%, var(--hairline));
1651
1656
  border-radius: var(--radius-lg);
1652
- background: color-mix(in srgb, var(--amber-soft, #fff7d6) 60%, var(--surface));
1657
+ background: color-mix(
1658
+ in srgb,
1659
+ var(--amber-soft, #fff7d6) 60%,
1660
+ var(--surface)
1661
+ );
1653
1662
  color: var(--body);
1654
1663
  font-size: 12px;
1655
1664
  }
@@ -3050,7 +3059,11 @@ button.copy-command:focus-visible {
3050
3059
  }
3051
3060
  .dags-copy-btn:active {
3052
3061
  border-color: var(--hairline-strong, #cfcdc4);
3053
- background: color-mix(in srgb, var(--surface-muted, #f0efe9) 80%, var(--hairline, #e6e5e0));
3062
+ background: color-mix(
3063
+ in srgb,
3064
+ var(--surface-muted, #f0efe9) 80%,
3065
+ var(--hairline, #e6e5e0)
3066
+ );
3054
3067
  }
3055
3068
  /* Mouse focus must not leave residual orange chrome; orange is :focus-visible only. */
3056
3069
  .dags-copy-btn:focus:not(:focus-visible) {
@@ -3201,12 +3214,20 @@ button.copy-command:focus-visible {
3201
3214
  }
3202
3215
  .dags-page-num:hover:not(:disabled),
3203
3216
  .dags-page-nav:hover:not(:disabled) {
3204
- background: color-mix(in srgb, var(--orange-soft, #fff1ea) 70%, var(--surface-muted, #f0efe9));
3217
+ background: color-mix(
3218
+ in srgb,
3219
+ var(--orange-soft, #fff1ea) 70%,
3220
+ var(--surface-muted, #f0efe9)
3221
+ );
3205
3222
  color: var(--ink, #26251e);
3206
3223
  }
3207
3224
  .dags-page-num:active:not(:disabled),
3208
3225
  .dags-page-nav:active:not(:disabled) {
3209
- background: color-mix(in srgb, var(--orange, #f54e00) 14%, var(--surface-muted, #f0efe9));
3226
+ background: color-mix(
3227
+ in srgb,
3228
+ var(--orange, #f54e00) 14%,
3229
+ var(--surface-muted, #f0efe9)
3230
+ );
3210
3231
  }
3211
3232
  .dags-page-num:focus-visible,
3212
3233
  .dags-page-nav:focus-visible,
@@ -3224,7 +3245,8 @@ button.copy-command:focus-visible {
3224
3245
  border-color: var(--orange, #f54e00);
3225
3246
  color: #fff;
3226
3247
  opacity: 1;
3227
- box-shadow: 0 2px 8px color-mix(in srgb, var(--orange, #f54e00) 35%, transparent);
3248
+ box-shadow: 0 2px 8px
3249
+ color-mix(in srgb, var(--orange, #f54e00) 35%, transparent);
3228
3250
  cursor: default;
3229
3251
  }
3230
3252
  .dags-page-nav:disabled,
@@ -3320,7 +3342,11 @@ button.copy-command:focus-visible {
3320
3342
  background: var(--surface-muted, #f0efe9);
3321
3343
  }
3322
3344
  .dags-refresh-btn:active:not(:disabled) {
3323
- background: color-mix(in srgb, var(--surface-muted, #f0efe9) 80%, var(--hairline, #e6e5e0));
3345
+ background: color-mix(
3346
+ in srgb,
3347
+ var(--surface-muted, #f0efe9) 80%,
3348
+ var(--hairline, #e6e5e0)
3349
+ );
3324
3350
  }
3325
3351
  .dags-pagination.is-busy {
3326
3352
  opacity: 0.92;
@@ -3516,7 +3542,11 @@ button.copy-command:focus-visible {
3516
3542
  }
3517
3543
  .custom-select-option.is-selected.is-active,
3518
3544
  .custom-select-option.is-selected:hover {
3519
- background: color-mix(in srgb, var(--orange-soft, #fff1ea) 70%, var(--surface-muted, #f0efe9));
3545
+ background: color-mix(
3546
+ in srgb,
3547
+ var(--orange-soft, #fff1ea) 70%,
3548
+ var(--surface-muted, #f0efe9)
3549
+ );
3520
3550
  }
3521
3551
  .custom-select-option-label {
3522
3552
  flex: 1 1 auto;
@@ -909,10 +909,6 @@ function symbolCaseId(symbol) {
909
909
  const match = symbol.match(/^test_(BE(?:_[A-Z0-9]+)+?_\d{2,3})(?:_|$)/i);
910
910
  return match ? canonicalCaseId(match[1].replaceAll("_", "-")) : undefined;
911
911
  }
912
- function metadataCaseIds(region) {
913
- return orderedUnique([...region.matchAll(/^\s*Case-ID\s*:\s*(BE-[A-Z0-9]+(?:-[A-Z0-9]+)*)\s*$/gmi)]
914
- .map((match) => canonicalCaseId(match[1])));
915
- }
916
912
  function metadataTestPoints(region, label) {
917
913
  const escaped = label.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
918
914
  const match = new RegExp(`^\\s*${escaped}\\s*:\\s*(.*)$`, "mi").exec(region);
@@ -932,27 +928,6 @@ function pytestFunctionRegionStart(source, functionIndex) {
932
928
  }
933
929
  return regionStart;
934
930
  }
935
- function pytestFunctionBodyEnd(source, signatureEnd, functionIndent) {
936
- let cursor = source.indexOf("\n", signatureEnd);
937
- if (cursor < 0)
938
- return source.length;
939
- cursor += 1;
940
- const baseIndent = functionIndent.replaceAll("\t", " ").length;
941
- while (cursor < source.length) {
942
- const nextLine = source.indexOf("\n", cursor);
943
- const lineEnd = nextLine < 0 ? source.length : nextLine;
944
- const line = source.slice(cursor, lineEnd).replace(/\r$/, "");
945
- if (line.trim()) {
946
- const indentation = line.match(/^[ \t]*/)?.[0]?.replaceAll("\t", " ").length ?? 0;
947
- if (indentation <= baseIndent)
948
- return cursor;
949
- }
950
- if (nextLine < 0)
951
- return source.length;
952
- cursor = nextLine + 1;
953
- }
954
- return source.length;
955
- }
956
931
  function pytestParameterCollections(source) {
957
932
  const collections = new Map();
958
933
  for (const match of source.matchAll(/^([A-Z][A-Z0-9_]*)\s*=\s*\[([\s\S]*?)^\]/gm)) {
@@ -966,10 +941,10 @@ function pytestSymbols(script, source) {
966
941
  const parameterCollections = pytestParameterCollections(source);
967
942
  return matches.map((match, index) => {
968
943
  const regionStart = regionStarts[index];
969
- const regionEnd = pytestFunctionBodyEnd(source, match.index + match[0].length, match[1]);
944
+ const regionEnd = regionStarts[index + 1] ?? source.length;
970
945
  const region = source.slice(match.index, regionEnd);
971
946
  const decorators = source.slice(regionStart, match.index);
972
- const ids = metadataCaseIds(region);
947
+ const ids = caseIds(region);
973
948
  const fromSymbol = symbolCaseId(match[2]);
974
949
  if (fromSymbol)
975
950
  ids.unshift(fromSymbol);
@@ -1,7 +1,5 @@
1
1
  import path from "node:path";
2
2
  const PRIORITY_ONLY_MODULE_STEMS = new Set(["p0", "p1", "p2"]);
3
- const OPAQUE_HASH_MODULE_STEM = /^[a-f][a-f0-9]{6,63}$/;
4
- export const BACKEND_TEST_MAX_MODULE_COUNT = 8;
5
3
  export function normalizeBackendTestModuleStemCandidate(raw) {
6
4
  return path.basename(raw)
7
5
  .replace(/\.(?:md|py)$/i, "")
@@ -13,9 +11,6 @@ export function normalizeBackendTestModuleStemCandidate(raw) {
13
11
  export function isPriorityOnlyBackendTestModuleStem(raw) {
14
12
  return PRIORITY_ONLY_MODULE_STEMS.has(normalizeBackendTestModuleStemCandidate(raw));
15
13
  }
16
- export function isOpaqueHashBackendTestModuleStem(raw) {
17
- return OPAQUE_HASH_MODULE_STEM.test(normalizeBackendTestModuleStemCandidate(raw));
18
- }
19
14
  export function isPriorityOnlyBackendPytestScript(script) {
20
15
  const basename = path.posix.basename(script.replaceAll("\\", "/"));
21
16
  const match = /^test_(.+)\.py$/i.exec(basename);
@@ -2,7 +2,7 @@ import { createHash } from "node:crypto";
2
2
  import { access, mkdir, readFile, readdir, writeFile } from "node:fs/promises";
3
3
  import path from "node:path";
4
4
  import { z } from "zod";
5
- import { BACKEND_TEST_MAX_MODULE_COUNT, isOpaqueHashBackendTestModuleStem, isPriorityOnlyBackendTestModuleStem, } from "./backend-test-module-stem.js";
5
+ import { isPriorityOnlyBackendTestModuleStem } from "./backend-test-module-stem.js";
6
6
  /**
7
7
  * Maps a backend-test generation writer task id to its writer-progress role.
8
8
  * Returns undefined for nodes that are not completeness-gated generators.
@@ -103,8 +103,6 @@ function moduleStemInvalidReason(raw) {
103
103
  const stem = normalizeBackendTestModuleStem(raw);
104
104
  if (isPriorityOnlyBackendTestModuleStem(stem))
105
105
  return "priority-only-module-stem";
106
- if (isOpaqueHashBackendTestModuleStem(stem))
107
- return "opaque-hash-module-stem";
108
106
  if (stem.toLowerCase() === "readme")
109
107
  return "reserved-module-stem";
110
108
  if (!/^[a-z][a-z0-9_]*$/.test(stem))
@@ -116,40 +114,28 @@ function moduleStemInvalidReason(raw) {
116
114
  function looksLikeValidModuleStem(raw) {
117
115
  return moduleStemInvalidReason(raw) === undefined;
118
116
  }
119
- function extractCanonicalModuleIndexSection(readme) {
120
- const lines = readme
121
- .replaceAll("\r\n", "\n")
122
- .replaceAll("\r", "\n")
123
- .split("\n");
124
- const headings = lines.flatMap((line, index) => line.trim() === "## Module Index" ? [index] : []);
125
- if (headings.length === 0) {
126
- return { section: "", structureIssues: ["missing-module-index"] };
127
- }
128
- if (headings.length > 1) {
129
- return { section: "", structureIssues: ["duplicate-module-index"] };
130
- }
131
- const start = headings[0] + 1;
132
- const relativeEnd = lines.slice(start).findIndex((line) => /^##\s+\S/.test(line.trim()));
133
- const end = relativeEnd < 0 ? lines.length : start + relativeEnd;
134
- return { section: lines.slice(start, end).join("\n"), structureIssues: [] };
135
- }
136
- /** Inspect only candidates inside the unique exact `## Module Index` section. */
117
+ /** Inspect only authoritative README module-index candidates. */
137
118
  export function inspectModuleStemsFromReadme(readme) {
138
- const extracted = extractCanonicalModuleIndexSection(readme);
139
119
  const candidates = [];
140
- const tableRowLines = extracted.section.split("\n").filter((line) => line.includes("|"));
120
+ const tableRowLines = readme
121
+ .replaceAll("\r\n", "\n")
122
+ .replaceAll("\r", "\n")
123
+ .split("\n")
124
+ .filter((line) => line.includes("|"));
141
125
  for (const line of tableRowLines) {
142
126
  for (const match of line.matchAll(/`?testcase\/md\/([A-Za-z0-9_.-]+)\.md`?/g)) {
143
127
  candidates.push(match[1]);
144
128
  }
145
129
  for (const match of line.matchAll(/\|\s*`?([A-Za-z0-9_.-]+)`?\s*\|\s*`?testcase\/test_/g)) {
146
- const raw = match[1];
147
- if (looksLikeValidModuleStem(raw) || moduleStemInvalidReason(raw) !== "invalid-syntax") {
148
- candidates.push(raw);
130
+ // This compatibility shape can also capture a numeric count column.
131
+ // Keep valid stems and the priority-only defect we diagnose, while
132
+ // ignoring non-stem incidental cells such as `1`.
133
+ if (looksLikeValidModuleStem(match[1]) || isPriorityOnlyBackendTestModuleStem(match[1])) {
134
+ candidates.push(match[1]);
149
135
  }
150
136
  }
151
137
  }
152
- for (const match of extracted.section.matchAll(/\[[^\]]+\]\(\.\/([A-Za-z0-9_.-]+)\.md\)/g)) {
138
+ for (const match of readme.matchAll(/\[[^\]]+\]\(\.\/([A-Za-z0-9_.-]+)\.md\)/g)) {
153
139
  candidates.push(match[1]);
154
140
  }
155
141
  const uniqueCandidates = orderedUnique(candidates);
@@ -159,18 +145,12 @@ export function inspectModuleStemsFromReadme(readme) {
159
145
  ? [{ raw, normalized: normalizeBackendTestModuleStem(raw), reasonCode }]
160
146
  : [];
161
147
  });
162
- const validStems = orderedUnique(uniqueCandidates
163
- .filter(looksLikeValidModuleStem)
164
- .map((stem) => normalizeBackendTestModuleStem(stem)));
165
- const structureIssues = [...extracted.structureIssues];
166
- if (validStems.length > BACKEND_TEST_MAX_MODULE_COUNT) {
167
- structureIssues.push(`excessive-module-count: ${validStems.length} > ${BACKEND_TEST_MAX_MODULE_COUNT}`);
168
- }
169
148
  return {
170
149
  candidates: uniqueCandidates,
171
- validStems,
150
+ validStems: orderedUnique(uniqueCandidates
151
+ .filter(looksLikeValidModuleStem)
152
+ .map((stem) => normalizeBackendTestModuleStem(stem))),
172
153
  invalid,
173
- structureIssues,
174
154
  };
175
155
  }
176
156
  /**
@@ -937,17 +917,6 @@ export async function assessBackendTestMdPlanCompleteness(workspaceRoot) {
937
917
  // are produced by map_agent children and are NOT required here.
938
918
  const stemInspection = inspectModuleStemsFromReadme(readme);
939
919
  const indexedStems = stemInspection.validStems;
940
- for (const structureIssue of stemInspection.structureIssues) {
941
- if (!brokenPaths.includes("testcase/md/README.md")) {
942
- brokenPaths.push("testcase/md/README.md");
943
- }
944
- issues.push({
945
- code: "T5",
946
- path: "testcase/md/README.md",
947
- detail: structureIssue,
948
- recoverable: true,
949
- });
950
- }
951
920
  for (const invalid of stemInspection.invalid) {
952
921
  if (!brokenPaths.includes("testcase/md/README.md")) {
953
922
  brokenPaths.push("testcase/md/README.md");
@@ -367,26 +367,19 @@ export async function executeDynamicMapExpansion(input) {
367
367
  }
368
368
  }
369
369
  const budgetBlockedIds = new Set(blockedChildren.map((entry) => entry.nodeId));
370
- const failOnTokenBudgetExhaustion = input.expansion.failOnTokenBudgetExhaustion === true;
371
370
  const failedChildren = tolerateChildFailures
372
371
  ? []
373
372
  : childNodeIds.filter((nodeId) => {
374
373
  const record = input.state.nodes[nodeId];
375
374
  if (!record || record.status === "FINISHED")
376
375
  return false;
377
- if (budgetBlockedIds.has(nodeId) || isBudgetBlockedRecord(record)) {
378
- return failOnTokenBudgetExhaustion;
379
- }
376
+ if (budgetBlockedIds.has(nodeId) || isBudgetBlockedRecord(record))
377
+ return false;
380
378
  return true;
381
379
  });
382
- const finishedItemCount = childNodeIds.filter((nodeId) => input.state.nodes[nodeId]?.status === "FINISHED").length;
383
- const blockedItemCount = budgetBlockedIds.size;
384
380
  const aggregate = {
385
381
  workflowNodeId: input.expansion.workflowNodeId,
386
382
  itemCount: items.length,
387
- expectedItemCount: items.length,
388
- finishedItemCount,
389
- blockedItemCount,
390
383
  children: childNodeIds.map((nodeId, index) => ({
391
384
  nodeId,
392
385
  item: items[index],
@@ -406,30 +399,21 @@ export async function executeDynamicMapExpansion(input) {
406
399
  ?.reason,
407
400
  })),
408
401
  tokensUsed: totalTokensUsed,
409
- maxTotalTokens: tokenBudget?.maxTotalTokens,
410
402
  caseOutcomeNotes: caseOutcomeNotes.length > 0 ? caseOutcomeNotes : undefined,
411
403
  };
412
- const tokenBudgetFailure = failOnTokenBudgetExhaustion && blockedItemCount > 0;
413
404
  return {
414
- // With tolerateChildFailures: barrier succeeds and product outcomes live in
415
- // case evidence. Backend-test opts into token-budget fail-closed because a
416
- // partial generated module set is never a valid downstream input.
417
- ok: failedChildren.length === 0 && !tokenBudgetFailure,
405
+ // With tolerateChildFailures: barrier always succeeds; product outcomes live in
406
+ // case evidence + result materialization. Without it: real child failures fail-close.
407
+ ok: failedChildren.length === 0,
418
408
  stdout: JSON.stringify(aggregate),
419
409
  stderr: caseOutcomeNotes.length > 0
420
410
  ? `map children recorded as case outcomes: ${caseOutcomeNotes
421
411
  .map((entry) => `${entry.nodeId}=${entry.status}:${entry.reason}`)
422
412
  .join(", ")}`
423
- : tokenBudgetFailure
424
- ? `dynamic map token budget exhausted: expected=${items.length} finished=${finishedItemCount} blocked=${blockedItemCount} tokensUsed=${totalTokensUsed} maxTotalTokens=${tokenBudget?.maxTotalTokens ?? "unbounded"}`
425
- : failedChildren.length > 0
426
- ? `dynamic map children failed: ${failedChildren.join(", ")}`
427
- : "",
428
- failureCategory: tokenBudgetFailure
429
- ? "dynamic-expansion-token-budget-exhausted"
430
413
  : failedChildren.length > 0
431
- ? "dynamic-expansion-child-failed"
432
- : "success",
414
+ ? `dynamic map children failed: ${failedChildren.join(", ")}`
415
+ : "",
416
+ failureCategory: failedChildren.length > 0 ? "dynamic-expansion-child-failed" : "success",
433
417
  durationMs: Date.now() - started,
434
418
  };
435
419
  }
@@ -1,31 +1,14 @@
1
1
  import { createHash } from "node:crypto";
2
- import { spawn } from "node:child_process";
3
2
  import { mkdir, readFile, writeFile } from "node:fs/promises";
4
3
  import path from "node:path";
4
+ import { runResilientGitCommand } from "../../shared/resilient-git.js";
5
5
  import { writeDagRunJsonArtifact } from "../../infrastructure/harness/artifact-store.js";
6
6
  import { pathMatchesPattern } from "../../shared/git-progress.js";
7
7
  export const FRONTEND_WORKTREE_DIFF_SCHEMA_ID = "frontend-worktree-diff-v1";
8
8
  export const FRONTEND_WORKTREE_BASELINE_SCHEMA_ID = "frontend-worktree-baseline-v1";
9
- function runGit(cwd, args) {
10
- return new Promise((resolve, reject) => {
11
- const child = spawn("git", args, {
12
- cwd,
13
- env: process.env,
14
- stdio: ["ignore", "pipe", "pipe"],
15
- });
16
- let stdout = "";
17
- let stderr = "";
18
- child.stdout.on("data", (chunk) => {
19
- stdout += String(chunk);
20
- });
21
- child.stderr.on("data", (chunk) => {
22
- stderr += String(chunk);
23
- });
24
- child.on("error", reject);
25
- child.on("close", (code) => {
26
- resolve({ code: code ?? 1, stdout, stderr });
27
- });
28
- });
9
+ async function runGit(cwd, args) {
10
+ const result = await runResilientGitCommand({ cwd, args, readOnly: true });
11
+ return { code: result.code, stdout: result.stdout, stderr: result.stderr };
29
12
  }
30
13
  function splitLines(text) {
31
14
  return text
@@ -52,8 +35,9 @@ async function sha256File(filePath) {
52
35
  }
53
36
  }
54
37
  export async function captureFrontendWorktreeBaseline(input) {
38
+ const run = input.dependencies?.runGit ?? runGit;
55
39
  const root = path.resolve(input.workspaceRoot);
56
- const status = await runGit(root, [
40
+ const status = await run(root, [
57
41
  "status",
58
42
  "--porcelain=v1",
59
43
  "--untracked-files=all",
@@ -84,6 +68,7 @@ export async function captureFrontendWorktreeBaseline(input) {
84
68
  * patch + inventory under the current run directory before review runs.
85
69
  */
86
70
  export async function runFrontendWorktreeDiffGate(input) {
71
+ const run = input.dependencies?.runGit ?? runGit;
87
72
  const root = path.resolve(input.workspaceRoot);
88
73
  let baseline;
89
74
  try {
@@ -101,16 +86,16 @@ export async function runFrontendWorktreeDiffGate(input) {
101
86
  if (input.requireBaseline && !baseline) {
102
87
  throw new Error("frontend worktree baseline is required for this review context");
103
88
  }
104
- const revParse = await runGit(root, ["rev-parse", "--is-inside-work-tree"]);
89
+ const revParse = await run(root, ["rev-parse", "--is-inside-work-tree"]);
105
90
  if (revParse.code !== 0 || revParse.stdout.trim() !== "true") {
106
91
  throw new Error(`frontend worktree diff gate requires a git worktree: ${revParse.stderr.trim() || revParse.stdout.trim() || "not a git repository"}`);
107
92
  }
108
93
  // Match failure capture semantics: binary-capable tracked diff vs HEAD.
109
- const diff = await runGit(root, ["diff", "--binary", "HEAD"]);
94
+ const diff = await run(root, ["diff", "--binary", "HEAD"]);
110
95
  if (diff.code !== 0) {
111
96
  throw new Error(`frontend worktree diff gate git diff failed: ${diff.stderr.trim() || diff.stdout.trim()}`);
112
97
  }
113
- const nameOnly = await runGit(root, [
98
+ const nameOnly = await run(root, [
114
99
  "diff",
115
100
  "--name-only",
116
101
  "--diff-filter=ACDMRTUXB",
@@ -119,7 +104,7 @@ export async function runFrontendWorktreeDiffGate(input) {
119
104
  if (nameOnly.code !== 0) {
120
105
  throw new Error(`frontend worktree diff gate git name-only failed: ${nameOnly.stderr.trim() || nameOnly.stdout.trim()}`);
121
106
  }
122
- const untracked = await runGit(root, [
107
+ const untracked = await run(root, [
123
108
  "ls-files",
124
109
  "--others",
125
110
  "--exclude-standard",
@@ -152,7 +137,7 @@ export async function runFrontendWorktreeDiffGate(input) {
152
137
  });
153
138
  const patchSource = baseline
154
139
  ? changedFiles.length > 0
155
- ? await runGit(root, ["diff", "--binary", "HEAD", "--", ...changedFiles])
140
+ ? await run(root, ["diff", "--binary", "HEAD", "--", ...changedFiles])
156
141
  : { code: 0, stdout: "", stderr: "" }
157
142
  : diff;
158
143
  if (patchSource.code !== 0) {