@tea-agent/loop-agent 0.26.0 → 0.26.1

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 (170) hide show
  1. package/CHANGELOG.md +1032 -1020
  2. package/README.md +8 -3
  3. package/bin/loop-agent.js +21 -21
  4. package/dist/cli/command-definitions.js +25 -10
  5. package/dist/cli/help.js +4 -3
  6. package/dist/cli/program.js +43 -17
  7. package/dist/commands/cursor-prompt.js +6 -6
  8. package/dist/commands/import-prd.js +7 -2
  9. package/dist/commands/init.js +7 -5
  10. package/dist/commands/loop-benchmark.js +11 -11
  11. package/dist/commands/pi-reuse-benchmark.js +16 -16
  12. package/dist/commands/task-source-prepare.js +468 -0
  13. package/dist/executors/dag-pi-executor.js +40 -5
  14. package/dist/executors/shell-write-guard.js +161 -25
  15. package/dist/sidecars/cursor-prompt/executor.js +1 -1
  16. package/dist/task/source-prepare/build-draft.js +215 -0
  17. package/dist/task/source-prepare/completeness.js +195 -0
  18. package/dist/task/source-prepare/index.js +7 -0
  19. package/dist/task/source-prepare/parse-intent.js +373 -0
  20. package/dist/task/source-prepare/path-policy.js +197 -0
  21. package/dist/task/source-prepare/prepare.js +506 -0
  22. package/dist/task/source-prepare/reference-integrity.js +274 -0
  23. package/dist/task/source-prepare/types.js +7 -0
  24. package/dist/worker/observe/static/copy.js +67 -67
  25. package/dist/worker/observe/static/dag-layout.d.ts +31 -31
  26. package/dist/worker/observe/static/dag-layout.js +83 -83
  27. package/dist/worker/observe/static/dom.js +220 -220
  28. package/dist/worker/observe/static/relations.js +133 -133
  29. package/dist/worker/observe/static/router.js +93 -93
  30. package/dist/worker/observe/static/run-processing.js +148 -148
  31. package/dist/worker/observe/static/views/batch.js +227 -227
  32. package/dist/worker/observe/static/views/dag-graph.js +172 -172
  33. package/dist/worker/observe/static/views/failures.js +143 -143
  34. package/dist/worker/observe/static/views/feature.js +492 -492
  35. package/dist/worker/observe/static/views/run.js +453 -453
  36. package/dist/worker/observe/static/views/shell.js +7 -7
  37. package/dist/worker/observe/static/views/timeline.js +163 -163
  38. package/dist/workflows/dag/canvas-observer.js +275 -275
  39. package/docs/skills/README.md +7 -7
  40. package/docs/templates/adr.md +60 -60
  41. package/docs/templates/agent-dag-authority-surface-audit.prompt.md +94 -94
  42. package/docs/templates/agent-dag-decision-envelope.schema.json +213 -213
  43. package/docs/templates/agent-dag-decision-gate.prompt.md +246 -246
  44. package/docs/templates/agent-dag-process-supervisor.prompt.md +98 -98
  45. package/docs/templates/agent-dag-report.schema.json +473 -473
  46. package/docs/templates/agent-dag-review-verdict.prompt.md +68 -68
  47. package/docs/templates/backend-test-result.schema.json +99 -99
  48. package/docs/templates/feature-spec.md +53 -53
  49. package/docs/templates/frontend-design-contract.md +42 -42
  50. package/docs/templates/frontend-eval/fixtures/failures/01-type-build-error.md +17 -17
  51. package/docs/templates/frontend-eval/fixtures/failures/02-unit-component-test-fail.md +16 -16
  52. package/docs/templates/frontend-eval/fixtures/failures/03-fixture-schema-drift.md +16 -16
  53. package/docs/templates/frontend-eval/fixtures/failures/04-missing-loading-empty-error-state.md +16 -16
  54. package/docs/templates/frontend-eval/fixtures/failures/05-forbidden-write-writeset-expansion.md +16 -16
  55. package/docs/templates/frontend-eval/fixtures/failures/06-unapproved-dependency-add.md +16 -16
  56. package/docs/templates/frontend-eval/fixtures/failures/07-mock-production-on.md +21 -21
  57. package/docs/templates/frontend-eval/fixtures/functional/01-simple-component-style.md +29 -29
  58. package/docs/templates/frontend-eval/fixtures/functional/02-form-validation.md +28 -28
  59. package/docs/templates/frontend-eval/fixtures/functional/03-list-detail-page.md +28 -28
  60. package/docs/templates/frontend-eval/fixtures/functional/04-api-mock.md +29 -29
  61. package/docs/templates/frontend-eval/fixtures/functional/05-permission-auth-gated-ui.md +27 -27
  62. package/docs/templates/frontend-eval/fixtures/functional/06-ssr-server-client-boundary.md +28 -28
  63. package/docs/templates/frontend-eval/fixtures/functional/07-shared-public-component-api.md +28 -28
  64. package/docs/templates/frontend-eval/fixtures/functional/08-pure-local-no-remote.md +27 -27
  65. package/docs/templates/frontend-eval/metrics.md +138 -138
  66. package/docs/templates/frontend-eval/smoke-targets.md +53 -53
  67. package/docs/templates/frontend-task-constraints.md +35 -35
  68. package/docs/templates/frontend-task-requirement.md +70 -70
  69. package/docs/templates/init-evolution-review.md +35 -35
  70. package/docs/templates/init-managed-agents.md +10 -5
  71. package/docs/templates/interactive-ui-round2-experiment.md +66 -66
  72. package/docs/templates/knowledge-graph-bootstrap-dag.json +118 -118
  73. package/docs/templates/knowledge-sync-dag.json +178 -178
  74. package/docs/templates/knowledge-sync-draft.schema.json +71 -71
  75. package/docs/templates/product-line/closeout.yaml +9 -9
  76. package/docs/templates/product-line/design.md +13 -13
  77. package/docs/templates/product-line/links.md +10 -10
  78. package/docs/templates/product-line/requirement.md +17 -17
  79. package/docs/templates/product-line/test-plan.md +7 -7
  80. package/docs/templates/production-readiness-checklist.md +57 -57
  81. package/docs/templates/project-start-checklist.md +9 -9
  82. package/docs/templates/qa-report.md +48 -48
  83. package/docs/templates/sprint-contract.md +29 -29
  84. package/docs/templates/worker-dogfood-evidence.md +80 -80
  85. package/docs/templates/worker-dogfood-setup.md +68 -68
  86. package/package.json +1 -1
  87. package/scripts/kb-bootstrap-init-skeleton.sh +0 -0
  88. package/scripts/kb-graph-incremental-prepare.mjs +386 -386
  89. package/scripts/kb-graph-materialize.mjs +105 -105
  90. package/scripts/kb-graph-promote.mjs +164 -164
  91. package/scripts/kb-query.mjs +554 -554
  92. package/skills/ai-engineering-context/SKILL.md +48 -48
  93. package/skills/analyze-product-dependencies/SKILL.md +67 -67
  94. package/skills/analyze-product-dependencies/agents/openai.yaml +4 -4
  95. package/skills/analyze-product-dependencies/references/api-documentation-schema.md +30 -30
  96. package/skills/analyze-product-dependencies/references/dependency-analysis-schema.md +28 -28
  97. package/skills/analyze-product-dependencies/references/example.md +76 -76
  98. package/skills/analyze-product-dependencies/references/forward-test-cases.md +35 -35
  99. package/skills/analyze-product-dependencies/references/input-contract.md +11 -11
  100. package/skills/analyze-product-dependencies/references/scouting-rules.md +61 -61
  101. package/skills/analyze-product-dependencies/scripts/test-validators.mjs +267 -267
  102. package/skills/analyze-product-dependencies/scripts/validate-api-documentation.mjs +101 -101
  103. package/skills/analyze-product-dependencies/scripts/validate-dependency-analysis.mjs +142 -142
  104. package/skills/analyze-product-dependencies/scripts/validate-product-requirement-input.mjs +76 -76
  105. package/skills/analyze-product-dependencies/scripts/validation-helpers.mjs +146 -146
  106. package/skills/analyze-product-requirements/SKILL.md +90 -90
  107. package/skills/analyze-product-requirements/agents/openai.yaml +4 -4
  108. package/skills/analyze-product-requirements/references/acceptance-criteria.md +91 -91
  109. package/skills/analyze-product-requirements/references/clarification-and-knowledge.md +56 -56
  110. package/skills/analyze-product-requirements/references/example.md +86 -86
  111. package/skills/analyze-product-requirements/references/forward-test-cases.md +66 -66
  112. package/skills/analyze-product-requirements/references/product-analysis-schema.md +32 -32
  113. package/skills/analyze-product-requirements/references/product-requirement-schema.md +33 -33
  114. package/skills/analyze-product-requirements/references/requirement-clarification-schema.md +35 -35
  115. package/skills/analyze-product-requirements/scripts/test-validators.mjs +193 -193
  116. package/skills/analyze-product-requirements/scripts/validate-product-analysis.mjs +69 -69
  117. package/skills/analyze-product-requirements/scripts/validate-product-requirement.mjs +97 -97
  118. package/skills/analyze-product-requirements/scripts/validate-requirement-clarification.mjs +98 -98
  119. package/skills/analyze-product-requirements/scripts/validation-helpers.mjs +156 -156
  120. package/skills/browser-tools/browser-content.js +103 -103
  121. package/skills/browser-tools/browser-cookies.js +35 -35
  122. package/skills/browser-tools/browser-eval.js +53 -53
  123. package/skills/browser-tools/browser-hn-scraper.js +108 -108
  124. package/skills/browser-tools/browser-nav.js +44 -44
  125. package/skills/browser-tools/browser-pick.js +162 -162
  126. package/skills/browser-tools/browser-screenshot.js +34 -34
  127. package/skills/browser-tools/browser-start.js +86 -86
  128. package/skills/browser-tools/package-lock.json +2556 -2556
  129. package/skills/browser-tools/package.json +19 -19
  130. package/skills/code-review-core/SKILL.md +20 -20
  131. package/skills/codebase-scout/SKILL.md +19 -19
  132. package/skills/grill-me/SKILL.md +10 -10
  133. package/skills/loop-agent/SKILL.md +5 -2
  134. package/skills/loop-agent/references/README.md +67 -67
  135. package/skills/loop-agent/references/command-reference.md +17 -15
  136. package/skills/loop-agent/references/docs-converge.md +126 -126
  137. package/skills/loop-agent/references/harness-policy.md +3 -4
  138. package/skills/loop-agent/references/learned/README.md +21 -21
  139. package/skills/loop-agent/references/long-running-loop.md +57 -57
  140. package/skills/loop-agent/references/one-shot-runs.md +85 -85
  141. package/skills/loop-agent/references/pi-prompt.md +23 -23
  142. package/skills/loop-agent/references/pi-subagent-assisted-mode.md +84 -84
  143. package/skills/loop-agent/references/post-implementation-and-patterns.md +1 -1
  144. package/skills/loop-agent/references/source-and-plan-practice.md +3 -2
  145. package/skills/loop-agent/references/task-workflow.md +7 -5
  146. package/skills/playwright-cli/SKILL.md +420 -420
  147. package/skills/playwright-cli/references/element-attributes.md +23 -23
  148. package/skills/playwright-cli/references/playwright-tests.md +39 -39
  149. package/skills/playwright-cli/references/request-mocking.md +87 -87
  150. package/skills/playwright-cli/references/running-code.md +241 -241
  151. package/skills/playwright-cli/references/session-management.md +225 -225
  152. package/skills/playwright-cli/references/storage-state.md +275 -275
  153. package/skills/playwright-cli/references/test-generation.md +433 -433
  154. package/skills/playwright-cli/references/tracing.md +139 -139
  155. package/skills/playwright-cli/references/video-recording.md +143 -143
  156. package/skills/requesting-code-review/SKILL.md +101 -101
  157. package/skills/requesting-code-review/code-reviewer.md +168 -168
  158. package/skills/systematic-debugging/CREATION-LOG.md +119 -119
  159. package/skills/systematic-debugging/condition-based-waiting-example.ts +158 -158
  160. package/skills/systematic-debugging/condition-based-waiting.md +115 -115
  161. package/skills/systematic-debugging/defense-in-depth.md +122 -122
  162. package/skills/systematic-debugging/find-polluter.sh +63 -63
  163. package/skills/systematic-debugging/root-cause-tracing.md +169 -169
  164. package/skills/systematic-debugging/test-academic.md +14 -14
  165. package/skills/systematic-debugging/test-pressure-1.md +58 -58
  166. package/skills/systematic-debugging/test-pressure-2.md +68 -68
  167. package/skills/systematic-debugging/test-pressure-3.md +69 -69
  168. package/skills/using-git-worktrees/SKILL.md +215 -215
  169. package/skills/verification-before-completion/SKILL.md +154 -154
  170. package/skills/webapp-testing/SKILL.md +19 -19
@@ -1,148 +1,148 @@
1
- /** Run step/command/artifact processing and task history rows. */
2
- import {
3
- STEP_LABELS,
4
- ARTIFACT_LABELS,
5
- STEP_ORDER,
6
- ARTIFACT_KEYS,
7
- } from "./constants.js";
8
- import { badge } from "./format.js";
9
- import { navigate } from "./router.js";
10
-
11
- export function stepDisplayLabel(label) {
12
- return STEP_LABELS[label] ?? label;
13
- }
14
-
15
- export function buildStepStates(events) {
16
- const states = new Map();
17
- for (const ev of events) {
18
- if (ev.type === "step.started") {
19
- states.set(ev.label, { status: "running", startedAt: ev.at });
20
- } else if (ev.type === "step.finished") {
21
- states.set(ev.label, {
22
- status: ev.status === "failed" ? "failed" : "succeeded",
23
- durationMs: ev.durationMs,
24
- });
25
- }
26
- }
27
-
28
- const ordered = [...STEP_ORDER];
29
- for (const label of states.keys()) {
30
- if (!ordered.includes(label)) ordered.push(label);
31
- }
32
-
33
- return ordered.map((label) => {
34
- const state = states.get(label);
35
- return {
36
- label,
37
- display: stepDisplayLabel(label),
38
- status: state?.status ?? "pending",
39
- durationMs: state?.durationMs,
40
- };
41
- });
42
- }
43
-
44
- export function stepMarker(status) {
45
- switch (status) {
46
- case "succeeded":
47
- return "✓";
48
- case "running":
49
- return "●";
50
- case "failed":
51
- return "✗";
52
- default:
53
- return "○";
54
- }
55
- }
56
-
57
- export function getCurrentCommand(events) {
58
- const active = new Map();
59
- for (const ev of events) {
60
- if (ev.type === "command.started") {
61
- active.set(ev.label, ev);
62
- } else if (ev.type === "command.finished") {
63
- active.delete(ev.label);
64
- }
65
- }
66
- let latest = null;
67
- for (const cmd of active.values()) {
68
- if (!latest || cmd.at > latest.at) latest = cmd;
69
- }
70
- if (latest?.at) {
71
- const elapsed = Date.now() - Date.parse(latest.at);
72
- latest = { ...latest, elapsedMs: Number.isFinite(elapsed) ? elapsed : 0 };
73
- }
74
- return latest;
75
- }
76
-
77
- export function getOutputTail(events, commandLabel) {
78
- if (!commandLabel) return "";
79
- const chunks = [];
80
- for (const ev of events) {
81
- if (
82
- ev.type === "command.output" &&
83
- ev.label === commandLabel &&
84
- ev.outputPreview
85
- ) {
86
- chunks.push(ev.outputPreview);
87
- }
88
- }
89
- return chunks.join("");
90
- }
91
-
92
- export function collectArtifacts(task, events) {
93
- const arts = new Map();
94
- const add = (label, artifactPath) => {
95
- if (artifactPath && !arts.has(label)) arts.set(label, artifactPath);
96
- };
97
-
98
- if (task?.artifactRefs) {
99
- for (const key of ARTIFACT_KEYS) {
100
- add(ARTIFACT_LABELS[key] ?? key, task.artifactRefs[key]);
101
- }
102
- }
103
-
104
- for (const ev of events) {
105
- if (!ev.artifactRefs) continue;
106
- for (const [key, artifactPath] of Object.entries(ev.artifactRefs)) {
107
- add(ARTIFACT_LABELS[key] ?? key, artifactPath);
108
- }
109
- }
110
-
111
- return arts;
112
- }
113
-
114
- export function findBatchRunIdForTask(snapshot, task) {
115
- if (!snapshot || !task) return null;
116
- const workerRunId = task.workerRunId;
117
- if (!workerRunId) return null;
118
- for (const batch of snapshot.batches ?? []) {
119
- for (const t of batch.tasks ?? []) {
120
- if (t.workerRunId === workerRunId) return batch.batchRunId;
121
- }
122
- }
123
- return null;
124
- }
125
-
126
- export function taskRunHistoryRow(run) {
127
- let runLink = run.workerRunId;
128
- if (run.workerRunId) {
129
- const a = document.createElement("a");
130
- a.href = `#/run/${encodeURIComponent(run.workerRunId)}`;
131
- a.textContent = run.workerRunId;
132
- a.addEventListener("click", (e) => {
133
- e.preventDefault();
134
- navigate(`#/run/${encodeURIComponent(run.workerRunId)}`);
135
- });
136
- runLink = a;
137
- }
138
- return {
139
- cells: [
140
- runLink,
141
- badge(run.status),
142
- run.recordedAt ?? run.finishedAt ?? "—",
143
- run.batchRunId ?? "—",
144
- run.retryOfWorkerRunId ?? "—",
145
- run.failureCategory ?? "—",
146
- ],
147
- };
148
- }
1
+ /** Run step/command/artifact processing and task history rows. */
2
+ import {
3
+ STEP_LABELS,
4
+ ARTIFACT_LABELS,
5
+ STEP_ORDER,
6
+ ARTIFACT_KEYS,
7
+ } from "./constants.js";
8
+ import { badge } from "./format.js";
9
+ import { navigate } from "./router.js";
10
+
11
+ export function stepDisplayLabel(label) {
12
+ return STEP_LABELS[label] ?? label;
13
+ }
14
+
15
+ export function buildStepStates(events) {
16
+ const states = new Map();
17
+ for (const ev of events) {
18
+ if (ev.type === "step.started") {
19
+ states.set(ev.label, { status: "running", startedAt: ev.at });
20
+ } else if (ev.type === "step.finished") {
21
+ states.set(ev.label, {
22
+ status: ev.status === "failed" ? "failed" : "succeeded",
23
+ durationMs: ev.durationMs,
24
+ });
25
+ }
26
+ }
27
+
28
+ const ordered = [...STEP_ORDER];
29
+ for (const label of states.keys()) {
30
+ if (!ordered.includes(label)) ordered.push(label);
31
+ }
32
+
33
+ return ordered.map((label) => {
34
+ const state = states.get(label);
35
+ return {
36
+ label,
37
+ display: stepDisplayLabel(label),
38
+ status: state?.status ?? "pending",
39
+ durationMs: state?.durationMs,
40
+ };
41
+ });
42
+ }
43
+
44
+ export function stepMarker(status) {
45
+ switch (status) {
46
+ case "succeeded":
47
+ return "✓";
48
+ case "running":
49
+ return "●";
50
+ case "failed":
51
+ return "✗";
52
+ default:
53
+ return "○";
54
+ }
55
+ }
56
+
57
+ export function getCurrentCommand(events) {
58
+ const active = new Map();
59
+ for (const ev of events) {
60
+ if (ev.type === "command.started") {
61
+ active.set(ev.label, ev);
62
+ } else if (ev.type === "command.finished") {
63
+ active.delete(ev.label);
64
+ }
65
+ }
66
+ let latest = null;
67
+ for (const cmd of active.values()) {
68
+ if (!latest || cmd.at > latest.at) latest = cmd;
69
+ }
70
+ if (latest?.at) {
71
+ const elapsed = Date.now() - Date.parse(latest.at);
72
+ latest = { ...latest, elapsedMs: Number.isFinite(elapsed) ? elapsed : 0 };
73
+ }
74
+ return latest;
75
+ }
76
+
77
+ export function getOutputTail(events, commandLabel) {
78
+ if (!commandLabel) return "";
79
+ const chunks = [];
80
+ for (const ev of events) {
81
+ if (
82
+ ev.type === "command.output" &&
83
+ ev.label === commandLabel &&
84
+ ev.outputPreview
85
+ ) {
86
+ chunks.push(ev.outputPreview);
87
+ }
88
+ }
89
+ return chunks.join("");
90
+ }
91
+
92
+ export function collectArtifacts(task, events) {
93
+ const arts = new Map();
94
+ const add = (label, artifactPath) => {
95
+ if (artifactPath && !arts.has(label)) arts.set(label, artifactPath);
96
+ };
97
+
98
+ if (task?.artifactRefs) {
99
+ for (const key of ARTIFACT_KEYS) {
100
+ add(ARTIFACT_LABELS[key] ?? key, task.artifactRefs[key]);
101
+ }
102
+ }
103
+
104
+ for (const ev of events) {
105
+ if (!ev.artifactRefs) continue;
106
+ for (const [key, artifactPath] of Object.entries(ev.artifactRefs)) {
107
+ add(ARTIFACT_LABELS[key] ?? key, artifactPath);
108
+ }
109
+ }
110
+
111
+ return arts;
112
+ }
113
+
114
+ export function findBatchRunIdForTask(snapshot, task) {
115
+ if (!snapshot || !task) return null;
116
+ const workerRunId = task.workerRunId;
117
+ if (!workerRunId) return null;
118
+ for (const batch of snapshot.batches ?? []) {
119
+ for (const t of batch.tasks ?? []) {
120
+ if (t.workerRunId === workerRunId) return batch.batchRunId;
121
+ }
122
+ }
123
+ return null;
124
+ }
125
+
126
+ export function taskRunHistoryRow(run) {
127
+ let runLink = run.workerRunId;
128
+ if (run.workerRunId) {
129
+ const a = document.createElement("a");
130
+ a.href = `#/run/${encodeURIComponent(run.workerRunId)}`;
131
+ a.textContent = run.workerRunId;
132
+ a.addEventListener("click", (e) => {
133
+ e.preventDefault();
134
+ navigate(`#/run/${encodeURIComponent(run.workerRunId)}`);
135
+ });
136
+ runLink = a;
137
+ }
138
+ return {
139
+ cells: [
140
+ runLink,
141
+ badge(run.status),
142
+ run.recordedAt ?? run.finishedAt ?? "—",
143
+ run.batchRunId ?? "—",
144
+ run.retryOfWorkerRunId ?? "—",
145
+ run.failureCategory ?? "—",
146
+ ],
147
+ };
148
+ }