owenloop 0.5.9 → 0.5.10

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 (83) hide show
  1. package/CHANGELOG.md +44 -0
  2. package/README.md +77 -37
  3. package/bin/owenloop.mjs +16 -1
  4. package/dist/packages/work/src/agent/loop.js +7 -0
  5. package/dist/packages/work/src/bundle/pull.d.ts +21 -0
  6. package/dist/packages/work/src/bundle/pull.js +118 -0
  7. package/dist/packages/work/src/exec/instructions.d.ts +4 -1
  8. package/dist/packages/work/src/exec/instructions.js +3 -0
  9. package/dist/packages/work/src/exec/loop.d.ts +1 -1
  10. package/dist/packages/work/src/exec/loop.js +92 -14
  11. package/dist/packages/work/src/harness/claude.js +16 -0
  12. package/dist/packages/work/src/harness/codex.d.ts +8 -0
  13. package/dist/packages/work/src/harness/codex.js +27 -1
  14. package/dist/packages/work/src/harness/contract.d.ts +9 -0
  15. package/dist/packages/work/src/roles/agent-run.js +38 -10
  16. package/dist/packages/work/src/roles/exec.js +25 -8
  17. package/dist/packages/work/src/shift/loop.js +118 -4
  18. package/dist/packages/work/src/shift/runtime.d.ts +3 -0
  19. package/dist/packages/work/src/shift/runtime.js +10 -1
  20. package/dist/packages/work/src/submit-proof.d.ts +11 -8
  21. package/dist/packages/work/src/submit-proof.js +11 -8
  22. package/dist/src/capabilities.d.ts +70 -6
  23. package/dist/src/capabilities.js +49 -6
  24. package/dist/src/capability-mapping-client.d.ts +29 -0
  25. package/dist/src/capability-mapping-client.js +160 -0
  26. package/dist/src/cli.d.ts +2 -2
  27. package/dist/src/cli.js +1196 -164
  28. package/dist/src/defs.d.ts +32 -1
  29. package/dist/src/defs.js +391 -79
  30. package/dist/src/engine.d.ts +62 -19
  31. package/dist/src/engine.js +180 -70
  32. package/dist/src/hub.d.ts +77 -0
  33. package/dist/src/hub.js +162 -0
  34. package/dist/src/implements.d.ts +56 -0
  35. package/dist/src/implements.js +790 -0
  36. package/dist/src/index.d.ts +8 -6
  37. package/dist/src/index.js +4 -3
  38. package/dist/src/mcp/serve.js +254 -45
  39. package/dist/src/mcp/server.d.ts +10 -7
  40. package/dist/src/mcp/server.js +53 -8
  41. package/dist/src/model.d.ts +46 -10
  42. package/dist/src/model.js +220 -74
  43. package/dist/src/store/def-source.d.ts +19 -0
  44. package/dist/src/store/def-source.js +87 -17
  45. package/dist/src/store/gc-recovery.d.ts +46 -0
  46. package/dist/src/store/gc-recovery.js +258 -0
  47. package/dist/src/store/gc.d.ts +94 -0
  48. package/dist/src/store/gc.js +562 -0
  49. package/dist/src/store/index.d.ts +3 -1
  50. package/dist/src/store/index.js +2 -0
  51. package/dist/src/store/install.d.ts +34 -5
  52. package/dist/src/store/install.js +73 -11
  53. package/dist/src/store/pre-commit-verifier.js +42 -2
  54. package/dist/src/store/snapshot-guard.d.ts +18 -0
  55. package/dist/src/store/snapshot-guard.js +94 -0
  56. package/dist/src/store.d.ts +64 -4
  57. package/dist/src/store.js +211 -8
  58. package/dist/src/types.d.ts +61 -17
  59. package/dist/src/util.js +4 -1
  60. package/docs/authoring.md +145 -8
  61. package/docs/bundles.md +19 -0
  62. package/docs/cli.md +270 -30
  63. package/docs/crypto.md +13 -7
  64. package/docs/design.md +107 -46
  65. package/docs/evals/mcp-charter-baseline.json +428 -0
  66. package/docs/mcp-charter-eval.md +124 -0
  67. package/docs/shift-logs.md +18 -0
  68. package/docs/wire-contracts.md +28 -7
  69. package/examples/workflows/compile-dev-playbook.yaml +1 -1
  70. package/examples/workflows/ship.yaml +28 -2
  71. package/package.json +2 -1
  72. package/plugins/claude-code/plugin/.claude-plugin/plugin.json +1 -1
  73. package/plugins/claude-code/plugin/.mcp.json +1 -1
  74. package/plugins/claude-code/plugin/skills/author/SKILL.md +22 -0
  75. package/plugins/claude-code/plugin/skills/ephemeral/SKILL.md +95 -0
  76. package/plugins/claude-code/plugin/skills/graduate/SKILL.md +156 -0
  77. package/plugins/claude-code/plugin/skills/plan/SKILL.md +236 -0
  78. package/plugins/codex/plugins/owenloop/.codex-plugin/plugin.json +1 -1
  79. package/plugins/codex/plugins/owenloop/.mcp.json +1 -1
  80. package/plugins/codex/plugins/owenloop/skills/author/SKILL.md +22 -0
  81. package/plugins/codex/plugins/owenloop/skills/ephemeral/SKILL.md +95 -0
  82. package/plugins/codex/plugins/owenloop/skills/graduate/SKILL.md +156 -0
  83. package/plugins/codex/plugins/owenloop/skills/plan/SKILL.md +236 -0
package/dist/src/store.js CHANGED
@@ -13,7 +13,125 @@
13
13
  * and (de)serialized at the boundary so callers always see real objects.
14
14
  */
15
15
  import { DatabaseSync } from 'node:sqlite';
16
+ import { lstatSync } from 'node:fs';
16
17
  import { detId, nowMs } from "./util.js";
18
+ import { compareStoreText, defDigest, isDefDigest, parseWorkflowCoordinate } from "./store/types.js";
19
+ import { withWorkflowSnapshotStoreGuard } from "./store/snapshot-guard.js";
20
+ function exactCallsFromSnapshot(record, workflow) {
21
+ const rawSteps = record.steps;
22
+ if (rawSteps === undefined)
23
+ return [];
24
+ if (!Array.isArray(rawSteps)) {
25
+ throw new Error(`workflow '${workflow}' has malformed def_snapshot.steps: expected an array`);
26
+ }
27
+ const exactCalls = new Set();
28
+ for (const [index, rawStep] of rawSteps.entries()) {
29
+ if (typeof rawStep !== 'object' || rawStep === null || Array.isArray(rawStep)) {
30
+ throw new Error(`workflow '${workflow}' has malformed def_snapshot.steps[${index}]: expected an object`);
31
+ }
32
+ const calls = rawStep.calls;
33
+ if (calls === undefined)
34
+ continue;
35
+ if (typeof calls !== 'string') {
36
+ throw new Error(`workflow '${workflow}' has malformed def_snapshot.steps[${index}].calls: expected a string`);
37
+ }
38
+ if (!calls.includes('@'))
39
+ continue;
40
+ try {
41
+ parseWorkflowCoordinate(calls);
42
+ }
43
+ catch (error) {
44
+ throw new Error(`workflow '${workflow}' has malformed exact def_snapshot calls target ${JSON.stringify(calls)}: ` +
45
+ error.message);
46
+ }
47
+ const rawLock = record.bundleLock;
48
+ if (typeof rawLock === 'object'
49
+ && rawLock !== null
50
+ && !Array.isArray(rawLock)
51
+ && Object.prototype.hasOwnProperty.call(rawLock, calls))
52
+ continue;
53
+ exactCalls.add(calls);
54
+ }
55
+ return [...exactCalls].sort(compareStoreText);
56
+ }
57
+ /**
58
+ * Read bundle pins from an EXISTING runtime database without migrating or
59
+ * otherwise writing it. GC deliberately cannot use {@link openStore}: opening
60
+ * the normal store enables write-oriented pragmas and schema migration, while
61
+ * a dry run must remain observational. Legacy databases without the workflow
62
+ * table or def_snapshot column simply predate snapshot pinning and contribute
63
+ * no pins.
64
+ */
65
+ export function readRuntimeSnapshotBundlePins(path) {
66
+ if (lstatSync(path, { throwIfNoEntry: false }) === undefined)
67
+ return [];
68
+ const db = new DatabaseSync(path, { readOnly: true });
69
+ try {
70
+ const hasMeta = db
71
+ .prepare(`SELECT 1 FROM sqlite_master WHERE type = 'table' AND name = 'meta'`)
72
+ .get() !== undefined;
73
+ if (hasMeta) {
74
+ const row = db.prepare(`SELECT v FROM meta WHERE k = 'schema_version'`).get();
75
+ const current = row?.v;
76
+ if (current !== undefined && parseInt(current, 10) > parseInt(SCHEMA_VERSION, 10)) {
77
+ throw new StoreVersionError(`database schema_version ${current} is newer than this owenloop's schema_version ${SCHEMA_VERSION}; ` +
78
+ 'upgrade your owenloop install to open this database');
79
+ }
80
+ }
81
+ const hasWorkflow = db
82
+ .prepare(`SELECT 1 FROM sqlite_master WHERE type = 'table' AND name = 'workflow'`)
83
+ .get() !== undefined;
84
+ if (!hasWorkflow)
85
+ return [];
86
+ const columns = db.prepare('PRAGMA table_info(workflow)').all();
87
+ if (!columns.some((column) => column.name === 'def_snapshot'))
88
+ return [];
89
+ const rows = db
90
+ .prepare('SELECT id, def_snapshot FROM workflow WHERE def_snapshot IS NOT NULL ORDER BY created_at, id')
91
+ .all();
92
+ return rows.map((row) => {
93
+ let snapshot;
94
+ try {
95
+ snapshot = JSON.parse(row.def_snapshot);
96
+ }
97
+ catch (error) {
98
+ throw new Error(`workflow '${row.id}' has malformed def_snapshot JSON: ${error.message}`);
99
+ }
100
+ if (typeof snapshot !== 'object' || snapshot === null || Array.isArray(snapshot)) {
101
+ throw new Error(`workflow '${row.id}' has malformed def_snapshot: expected an object`);
102
+ }
103
+ const record = snapshot;
104
+ const rawDigest = record.bundleDigest;
105
+ let bundleDigest;
106
+ if (rawDigest !== undefined && rawDigest !== '') {
107
+ if (typeof rawDigest !== 'string' || !isDefDigest(rawDigest)) {
108
+ throw new Error(`workflow '${row.id}' has noncanonical def_snapshot.bundleDigest ${JSON.stringify(rawDigest)}`);
109
+ }
110
+ bundleDigest = rawDigest;
111
+ }
112
+ const rawLock = record.bundleLock;
113
+ if (rawLock !== undefined && (typeof rawLock !== 'object' || rawLock === null || Array.isArray(rawLock))) {
114
+ throw new Error(`workflow '${row.id}' has malformed def_snapshot.bundleLock: expected an object`);
115
+ }
116
+ const bundleLock = Object.entries((rawLock ?? {})).map(([target, digest]) => {
117
+ if (typeof digest !== 'string' || !isDefDigest(digest)) {
118
+ throw new Error(`workflow '${row.id}' has noncanonical def_snapshot.bundleLock[${JSON.stringify(target)}] ` +
119
+ `${JSON.stringify(digest)}`);
120
+ }
121
+ return digest;
122
+ });
123
+ const exactCalls = exactCallsFromSnapshot(record, row.id);
124
+ return {
125
+ ...(bundleDigest === undefined ? {} : { bundleDigest }),
126
+ bundleLock: [...new Set(bundleLock)].sort(compareStoreText),
127
+ ...(exactCalls.length === 0 ? {} : { exactCalls }),
128
+ };
129
+ });
130
+ }
131
+ finally {
132
+ db.close();
133
+ }
134
+ }
17
135
  // ---- deterministic ids -------------------------------------------------------
18
136
  export function artifactId(workflow, path) {
19
137
  return detId('art', workflow, path);
@@ -385,6 +503,7 @@ function mapWorkflow(r) {
385
503
  // ---- the store ---------------------------------------------------------------
386
504
  export class Store {
387
505
  db;
506
+ activeSnapshotDigests = new Set();
388
507
  constructor(path) {
389
508
  this.db = new DatabaseSync(path);
390
509
  // Connection-scoped only — no file mutation, safe before the version check.
@@ -581,6 +700,35 @@ export class Store {
581
700
  throw err;
582
701
  }
583
702
  }
703
+ /**
704
+ * Persist one or more CAS-backed definition snapshots under the same
705
+ * workflow-store lock used by install and GC. Reachability is revalidated
706
+ * before `BEGIN IMMEDIATE`; the guarded digest set then authorizes only the
707
+ * low-level snapshot writes performed by this transaction.
708
+ */
709
+ txWithWorkflowSnapshots(snapshots, fn) {
710
+ const defs = Array.isArray(snapshots) ? snapshots : [snapshots];
711
+ return withWorkflowSnapshotStoreGuard(defs, (guardedDigests) => {
712
+ for (const digest of guardedDigests)
713
+ this.activeSnapshotDigests.add(digest);
714
+ try {
715
+ return this.tx(fn);
716
+ }
717
+ finally {
718
+ for (const digest of guardedDigests)
719
+ this.activeSnapshotDigests.delete(digest);
720
+ }
721
+ });
722
+ }
723
+ assertWorkflowSnapshotGuard(snapshot) {
724
+ if (snapshot?.bundleDigest === undefined || (snapshot.bundleStoreRoots?.length ?? 0) === 0)
725
+ return;
726
+ const digest = defDigest(snapshot.bundleDigest);
727
+ if (!this.activeSnapshotDigests.has(digest)) {
728
+ throw new Error(`refusing uncoordinated workflow snapshot write for bundle ${digest}: ` +
729
+ 'use txWithWorkflowSnapshots so bundle GC cannot race the commit');
730
+ }
731
+ }
584
732
  // -- meta --------------------------------------------------------------------
585
733
  getMeta(k) {
586
734
  const row = this.db.prepare('SELECT v FROM meta WHERE k = ?').get(k);
@@ -593,6 +741,7 @@ export class Store {
593
741
  }
594
742
  // -- workflow ----------------------------------------------------------------
595
743
  insertWorkflow(id, data, producedBy) {
744
+ this.assertWorkflowSnapshotGuard(data.defSnapshot);
596
745
  const at = nowMs();
597
746
  this.db
598
747
  .prepare(`INSERT INTO workflow
@@ -607,6 +756,7 @@ export class Store {
607
756
  * decide what "drift" means, it just persists what the engine computed.
608
757
  */
609
758
  repinWorkflowDef(id, snapshot, hash) {
759
+ this.assertWorkflowSnapshotGuard(snapshot);
610
760
  this.db
611
761
  .prepare('UPDATE workflow SET def_snapshot = ?, def_hash = ? WHERE id = ?')
612
762
  .run(JSON.stringify(snapshot), hash, id);
@@ -924,8 +1074,10 @@ export class Store {
924
1074
  if (!cur)
925
1075
  throw new Error(`run not found: ${id}`);
926
1076
  // order_json is DELIBERATELY excluded from `merged` and the UPDATE below:
927
- // the order packet is immutable after claim (§8 / Gap 1). Omitting it makes
928
- // that structural — no close/outcome/summary write can ever clobber it.
1077
+ // no close/outcome/summary write can ever clobber the order packet, and
1078
+ // omitting the column makes that structural rather than a convention.
1079
+ // `restampOrderTarget` below is the ONE writer of order_json after claim,
1080
+ // and it rewrites a single `owes[].version` and nothing else.
929
1081
  const merged = {
930
1082
  workflow: cur.workflow,
931
1083
  step: cur.step,
@@ -941,28 +1093,79 @@ export class Store {
941
1093
  .run(merged.key ?? '', merged.outcome ?? null, merged.summary ?? null, merged.sessionId ?? null, toJson(merged.fingerprint), merged.cause ?? null, nowMs(), id);
942
1094
  return this.getRun(id);
943
1095
  }
1096
+ /**
1097
+ * Re-stamp ONE owed version target on a run's persisted order packet.
1098
+ *
1099
+ * The order is frozen at claim with exactly one exception, and this is it.
1100
+ * When a reject re-arms a still-open claim, the artifact has already been
1101
+ * bumped by the commit that preceded the reject, so the target the claim
1102
+ * froze at issue names a version the next commit will overshoot — and the
1103
+ * producer would sign that stale number into its submission proof while the
1104
+ * hub keys the stored proof row by the version actually committed. Everything
1105
+ * else about the packet stays frozen, and `updateRun` still cannot touch it
1106
+ * at all.
1107
+ *
1108
+ * No-ops when the run has no persisted order (pre-schema-v7 rows), when the
1109
+ * order does not owe `path`, or when the target already is `version` — so
1110
+ * callers can call it unconditionally, and re-stamping is idempotent.
1111
+ */
1112
+ restampOrderTarget(id, path, version) {
1113
+ const cur = this.getRun(id);
1114
+ if (!cur?.order)
1115
+ return;
1116
+ const owed = cur.order.owes.find((o) => o.path === path);
1117
+ if (owed === undefined || owed.version === version)
1118
+ return;
1119
+ // Rebuilt immutably (map, not in-place mutation) so no caller is left
1120
+ // holding a half-mutated Order read before this write.
1121
+ const next = {
1122
+ ...cur.order,
1123
+ owes: cur.order.owes.map((o) => (o.path === path ? { ...o, version } : o)),
1124
+ };
1125
+ this.db
1126
+ .prepare('UPDATE run SET order_json = ?, updated_at = ? WHERE id = ?')
1127
+ .run(toJson(next), nowMs(), id);
1128
+ }
944
1129
  getRun(id) {
945
1130
  const r = this.db.prepare('SELECT * FROM run WHERE id = ?').get(id);
946
1131
  return r ? mapRun(r) : undefined;
947
1132
  }
948
- /** How many runs of this step since `sinceMs` (for the daily budget window). */
1133
+ /**
1134
+ * How many runs of this step since `sinceMs` (for the daily budget window).
1135
+ *
1136
+ * `released` RUNS ARE EXCLUDED — a lease returned unused is not a firing. See
1137
+ * `RunData.outcome` for why this is not the same as `no_work`, which stays
1138
+ * counted because the step did run.
1139
+ */
949
1140
  countRuns(workflow, step, sinceMs) {
950
1141
  const row = this.db
951
- .prepare('SELECT COUNT(*) AS n FROM run WHERE workflow = ? AND step = ? AND created_at >= ?')
1142
+ .prepare('SELECT COUNT(*) AS n FROM run WHERE workflow = ? AND step = ? AND created_at >= ?'
1143
+ + " AND (outcome IS NULL OR outcome != 'released')")
952
1144
  .get(workflow, step, sinceMs);
953
1145
  return row.n;
954
1146
  }
955
- /** The most recent run of this step, if any (for cadence gating). */
1147
+ /**
1148
+ * The most recent run of this step, if any (for cadence gating).
1149
+ *
1150
+ * `released` runs are excluded for the same reason `countRuns` excludes them:
1151
+ * a lease handed straight back did not fire the step, so it must not restart
1152
+ * the cadence clock. Without this a server that claims and releases on every
1153
+ * sweep holds a throttled step off indefinitely — each handback pushes the
1154
+ * next eligible time further out than the sweep interval that produced it.
1155
+ */
956
1156
  latestRun(workflow, step) {
957
1157
  const r = this.db
958
- .prepare('SELECT * FROM run WHERE workflow = ? AND step = ? ORDER BY created_at DESC LIMIT 1')
1158
+ .prepare("SELECT * FROM run WHERE workflow = ? AND step = ? AND (outcome IS NULL OR outcome != 'released')"
1159
+ + ' ORDER BY created_at DESC LIMIT 1')
959
1160
  .get(workflow, step);
960
1161
  return r ? mapRun(r) : undefined;
961
1162
  }
962
1163
  /**
963
1164
  * Count of consecutive trailing `failed` runs for this step+key — the
964
- * crash-step signal. Any closed run that is NOT `failed` (ok/no_work/skipped)
965
- * breaks the streak; still-open runs (outcome NULL) are ignored.
1165
+ * crash-step signal. Any closed run that is NOT `failed`
1166
+ * (ok/no_work/released/skipped) breaks the streak; still-open runs (outcome
1167
+ * NULL) are ignored. `released` breaks it exactly as `no_work` did before the
1168
+ * two were split, so the crash-step signal is unchanged by that split.
966
1169
  */
967
1170
  recentFailedRuns(workflow, step, key = '') {
968
1171
  const rows = this.db
@@ -170,6 +170,12 @@ export interface Order {
170
170
  * (`wise` → `wise:deep`), or the authored names verbatim when the run
171
171
  * carries no modifier. Absent when the step authored none.
172
172
  *
173
+ * A caller may also supply a capability MAPPING, which substitutes its own
174
+ * capability name for the authored one before the modifier is attached. Such
175
+ * an order carries no in-band marker — unlike a reroute, nothing here says a
176
+ * mapping fired. The authored name stays recoverable through `defDigest`,
177
+ * which pins the def that wrote it.
178
+ *
173
179
  * This is the value a claiming crew's bindings are matched against and the
174
180
  * value the shift's settings map is keyed by. It is a snapshot of the offer:
175
181
  * an already-claimed order is never recomposed, so re-reading it later
@@ -263,11 +269,12 @@ export interface Order {
263
269
  * stalls the step. This closes that: the requirement is stated up front,
264
270
  * from the same declaration the refusal will be measured against.
265
271
  *
266
- * It is a COPY, not a reference. The order is immutable once written
267
- * (`RunData.order`), so this is the schema as it stood at claim time; a def
268
- * edited mid-run leaves the projected schema and the enforced one able to
269
- * disagree, exactly as `defDigest` already records for the rest of the
270
- * order.
272
+ * It is a COPY, not a reference. Nothing rewrites the schema on a persisted
273
+ * order — `Store.restampOrderTarget`, the one post-claim writer, touches
274
+ * only `owes[].version` — so this is the schema as it stood at claim time;
275
+ * a def edited mid-run leaves the projected schema and the enforced one
276
+ * able to disagree, exactly as `defDigest` already records for the rest of
277
+ * the order.
271
278
  */
272
279
  schema?: JsonSchema;
273
280
  /**
@@ -299,7 +306,31 @@ export interface RunData {
299
306
  workflow: string;
300
307
  step: string;
301
308
  key?: string;
302
- outcome?: 'ok' | 'no_work' | 'failed' | 'skipped';
309
+ /**
310
+ * How the run ended, or `undefined` while it is still open.
311
+ *
312
+ * `no_work` and `released` are NOT the same thing, and the difference is
313
+ * load-bearing for the daily budget and the cadence gate:
314
+ *
315
+ * - `no_work` — the step RAN and had nothing to produce. A `merge-gate` that
316
+ * polled CI and found it still pending, an `ask` that handed the question
317
+ * to a human, a holder that rejected its input. Work happened; these runs
318
+ * spend budget and restart the cadence clock, which is exactly how a
319
+ * `cadenceSecs` poll stays throttled.
320
+ * - `released` — the LEASE WAS RETURNED UNUSED. Nothing ran. A server already
321
+ * at its agent cap handing the claim straight back or a pickup window that
322
+ * lapsed with no first contact increments the task's lease-churn
323
+ * `attempts`, while a private born-reject re-arm does not. All remain
324
+ * outside the daily budget and cadence: they are not firings, they are
325
+ * claims that were withdrawn.
326
+ *
327
+ * Conflating them stalled a live run. Six capacity handbacks against a step
328
+ * declaring `maxRunsPerDay: 6` spent the whole allowance before it executed
329
+ * once; the engine then deferred it `daily-budget` until local midnight while
330
+ * its artifact still read `attemptsUsed: 0`. Lease-churn attempts are status
331
+ * telemetry, not the rework budget (`judgmentRejects`) and not a run.
332
+ */
333
+ outcome?: 'ok' | 'no_work' | 'released' | 'failed' | 'skipped';
303
334
  summary?: string;
304
335
  sessionId?: string;
305
336
  /** the version of every consumed input at claim time (§12.2 commit CAS) */
@@ -308,8 +339,10 @@ export interface RunData {
308
339
  cause?: FiringTrigger;
309
340
  /** The flattened order packet issued at claim time (§8 / Gap 1) — the exact
310
341
  * Order buildOrder emitted, written in the SAME transaction that created
311
- * this run. Immutable after insert: updateRun never touches it. The
312
- * replay/eval/paper-trail record (buildOrder is deterministic modulo run
342
+ * this run. `updateRun` never touches it, so no close/outcome write can
343
+ * clobber it; `Store.restampOrderTarget` is the single narrow exception,
344
+ * rewriting one `owes[].version` when a reject re-arms a still-open claim.
345
+ * The replay/eval/paper-trail record (buildOrder is deterministic modulo run
313
346
  * id). Absent on runs created before schema v7. */
314
347
  order?: Order;
315
348
  }
@@ -803,11 +836,20 @@ export interface CheckReport {
803
836
  bounded: boolean;
804
837
  /** Which bounds were hit, for honest reporting. */
805
838
  boundsHit: ('maxDepth' | 'maxStates')[];
839
+ /**
840
+ * True when the search expanded at least one reachable `emit-seal` outcome.
841
+ * This describes the search shape, not a BFS stop condition, and does not
842
+ * contribute to `bounded`.
843
+ */
844
+ collectionCapApplied: boolean;
845
+ /** Effective per-seal member-count cap, including the default when unspecified. */
846
+ maxCollectionSize: number;
806
847
  /**
807
848
  * Reachable non-done states with zero eligible firings that have NO path to
808
- * completion even at unlimited attempts — i.e. recomputing eligibility with
809
- * all freezes lifted (a human `retry` = unlimited attempts) STILL yields
810
- * zero firings. A genuine structural dead-end (TRUE deadlock). A definite
849
+ * completion even after unlimited attempts and elapsed idle time — i.e.
850
+ * recomputing eligibility with all freezes lifted (a human `retry` =
851
+ * unlimited attempts) and eventual idle time STILL yields zero firings. A
852
+ * genuine structural dead-end (TRUE deadlock). A definite
811
853
  * defect only when the search was exhaustive (`!bounded`) — the
812
854
  * maxCollectionSize cap can otherwise manufacture a spurious no-moves
813
855
  * state. Always present ([] when none).
@@ -815,12 +857,14 @@ export interface CheckReport {
815
857
  deadlocks: CheckFinding[];
816
858
  /**
817
859
  * Reachable non-done states with zero eligible firings whose ONLY blocker
818
- * is a frozen/stalled debt (maxAttempts / maxSchemaFailures / held).
819
- * Recomputing eligibility with all freezes lifted (a human `retry` =
820
- * unlimited attempts) yields >= 1 firing, so a producer would re-arm and
821
- * the line could move. These are the DESIGNED human-escalation brakes —
822
- * EXPECTED, never a defect; they do not affect the exit code. Always
823
- * present ([] when none).
860
+ * is a frozen/stalled debt (maxAttempts / maxSchemaFailures / held) or an
861
+ * idle trigger whose threshold has not yet elapsed. Recomputing eligibility
862
+ * with all freezes lifted (a human `retry` = unlimited attempts) and
863
+ * eventual idle time yields >= 1 firing, so a producer would re-arm or an
864
+ * idle step would become eligible. These are expected brakes/waits — never
865
+ * a defect; they do not affect the exit code. This classification does not
866
+ * itself make an idle successor reachable in the timeless BFS, so
867
+ * `completable` can remain false. Always present ([] when none).
824
868
  */
825
869
  stallStates: CheckFinding[];
826
870
  /**
package/dist/src/util.js CHANGED
@@ -138,7 +138,10 @@ export function parseDurationMs(spec) {
138
138
  const m = DUR_RE.exec(String(spec).trim());
139
139
  if (!m)
140
140
  throw new Error(`bad duration: ${spec} (use 90m, 2h, or seconds)`);
141
- return Number(m[1]) * UNIT_MS[m[2]];
141
+ const milliseconds = Number(m[1]) * UNIT_MS[m[2]];
142
+ if (!Number.isFinite(milliseconds))
143
+ throw new Error(`bad duration: ${spec} (duration is too large)`);
144
+ return milliseconds;
142
145
  }
143
146
  /** "30m" → 1800 seconds. */
144
147
  export function parseDurationSecs(spec) {
package/docs/authoring.md CHANGED
@@ -92,6 +92,92 @@ steps:
92
92
  anything: goes # untouched onto the order (Order.x); see design.md §27.3
93
93
  ```
94
94
 
95
+ ## x.discovery — advisory workflow discovery metadata
96
+
97
+ x.discovery is an Owenloop-owned convention inside the otherwise opaque
98
+ top-level x: map. It lets a person or catalog tool understand what a workflow
99
+ does, when to choose it, when a close-looking workflow is the wrong choice, and
100
+ how to seed/read its public interface:
101
+
102
+ ~~~yaml
103
+ x:
104
+ discovery:
105
+ description: >-
106
+ One non-empty paragraph explaining the workflow in colleague-facing language.
107
+ whenToUse:
108
+ - a non-empty trigger phrase
109
+ notFor:
110
+ - a non-empty anti-trigger phrase
111
+ interface:
112
+ inputs:
113
+ - name: proposal
114
+ summary: A non-empty description of the seed value.
115
+ schemaRef: "#/inputs/0/schema"
116
+ outputs:
117
+ - name: merge
118
+ summary: A non-empty description of the public result.
119
+ schemaRef: "#/steps/6/produces/0/schema"
120
+ ~~~
121
+
122
+ The bag itself is optional today, so existing definitions remain compatible.
123
+ Its absence is a lint warning, not a loading or execution failure. Once the bag
124
+ is present, all four top-level fields are required: catalog browsing
125
+ (description), positive routing (whenToUse), false-positive avoidance (notFor),
126
+ and usable seeding/result information (interface) answer separate questions. In
127
+ particular, an anti-trigger is not redundant with a trigger: it rules out the
128
+ nearby cases that would otherwise appear to match.
129
+
130
+ description must be a non-empty string. whenToUse and notFor must each be
131
+ non-empty arrays of non-empty strings. interface.inputs and interface.outputs
132
+ are required arrays, though either can be empty when the workflow genuinely has
133
+ no declared inputs or no public outputs. Each entry must have a non-empty name,
134
+ summary, and schemaRef. The entries cover every declared input and every public
135
+ outputs: name exactly once; no unknown or duplicate names are allowed. Their
136
+ named artifacts must carry schemas. schemaRef is a local JSON Pointer beginning
137
+ #/ to that existing machine schema, not a second copy of the schema.
138
+
139
+ owenloop lint reports missing or malformed discovery data as field-specific
140
+ warnings. These warnings do not fail lint, loading, packing, publishing, pushing,
141
+ or execution. Other x vocabularies remain opaque and unrestricted; only this
142
+ authoring-lint convention is recognized.
143
+
144
+ ## x.implements — advisory external interface claims
145
+
146
+ When present, x.implements must be a non-empty array of external interface
147
+ coordinates a definition claims to satisfy:
148
+
149
+ ~~~yaml
150
+ x:
151
+ implements:
152
+ - name: research-report
153
+ version: "1"
154
+ - name: auditable-report
155
+ version: "2026-08"
156
+ ~~~
157
+
158
+ Every entry contains exactly non-empty string name and version fields. More
159
+ than one claim is allowed, but an exact name/version coordinate may appear only
160
+ once. Version is intentionally an opaque string here: Owenloop does not impose
161
+ SemVer or catalog policy. Malformed claims are warning-only authoring lint, so
162
+ they neither stop lint nor turn loading or execution into a failure. Keeping
163
+ the field under x also means binaries at engine version 1 ignore it and continue
164
+ to load definitions written before or after the convention.
165
+
166
+ x.discovery.interface and x.implements answer different questions.
167
+ x.discovery.interface describes this definition's own inputs and public outputs
168
+ and owns the local schemaRef vocabulary. x.implements only names an external
169
+ catalog contract; it deliberately copies no schemas, artifact mappings, or
170
+ summaries and introduces no second pointer convention. Catalog lookup and
171
+ publish-time rejection are hub responsibilities, not owenloop lint.
172
+
173
+ Catalog or publish tooling that needs to compare a claimed interface with a
174
+ workflow can use the exported `checkInterfaceCompatibility` helper. It checks
175
+ inputs contravariantly and outputs covariantly using the supported structural
176
+ JSON Schema subset. When a schema combines root constraints with a top-level
177
+ `anyOf` or `oneOf`, those root constraints remain in force while branch coverage
178
+ is checked; union branches are not treated as standalone schemas. The checker
179
+ continues to fail closed for unsupported or unprovable schema constructs.
180
+
95
181
  ## `model:` — quality tiers, not vendor ids
96
182
 
97
183
  The engine never calls a model; `model:` is an opaque string that rides the
@@ -311,6 +397,52 @@ check.
311
397
  `capabilities:` says *which* caller may claim it in the first place — a tick filter
312
398
  locally, a bound crew on a hub.
313
399
 
400
+ ### The trust boundary: publishing shares, installing scopes
401
+
402
+ `push` and `install` both put a def on your hub, and they treat its capabilities
403
+ differently on purpose.
404
+
405
+ `owenloop push` publishes defs **your org authored**. Their capabilities join the
406
+ org's **shared** vocabulary as written: two of your own defs that both author
407
+ `review` deliberately mean the same `review`, served by the same crews. That is
408
+ the whole value of a shared vocabulary.
409
+
410
+ `owenloop install <owner>/<repo>` publishes a def **someone else authored**. An
411
+ outside author writing `review` is not making your org's claim, so every
412
+ capability an installed def authors becomes `<def-name>.<capability>` —
413
+ `analyzer.review`, not `review` — unless you say otherwise, once, at install
414
+ time. Scoping is the default because the failure it prevents is silent: an
415
+ outside def entering the shared vocabulary starts drawing orders from the crews
416
+ already bound to that name.
417
+
418
+ Link deliberately, when you actually mean it. If your `reviewers` crew genuinely
419
+ should serve the installed def's `review`, answer the install prompt with
420
+ `code-review` (or whatever your org calls it), or pass
421
+ `--map review=code-review`. Then the org name it takes is the one your crews are
422
+ already bound to, and it draws from them by design rather than by collision.
423
+
424
+ **Never edit a def's content to get the mapping you want.** The mapping is
425
+ org-side data, recorded against `(def, authored-name)` on your hub; the def stays
426
+ byte-for-byte as its author wrote it, so upstream updates keep applying. Editing
427
+ `capabilities:` in a def you did not author forks it.
428
+
429
+ Four names, in the order the engine derives them:
430
+
431
+ | Vocabulary | Where it comes from |
432
+ | --- | --- |
433
+ | **authored** | what the def's YAML says, e.g. `review` |
434
+ | **mapped** | the org name a capability mapping gives it, e.g. `analyzer.review` |
435
+ | **composed** | the mapped name plus the run's modifier, e.g. `analyzer.review:deep` |
436
+ | **offered** | the composed name after any capability reroute rewrites it |
437
+
438
+ The two rewrite points key on different halves and fire at different times.
439
+ **Mappings key on the AUTHORED name and apply BEFORE composition** — they answer
440
+ "what does this def's `review` mean here". **Reroutes key on the COMPOSED string
441
+ and apply AFTER composition** — they answer "where should `analyzer.review:deep`
442
+ go right now". A reroute is therefore the operational lever (move traffic
443
+ today); a mapping is the identity decision (what this def's capability *is* in
444
+ your org).
445
+
314
446
  ## `produces:` vs `generates:`
315
447
 
316
448
  A stem under `produces:` is expected to be consumed downstream — owenloop's lint warns
@@ -482,11 +614,13 @@ produces:
482
614
  - **`atLeastOne`** — no refusal, no auto-skip; once any one member is green,
483
615
  the rest no longer count as outstanding for done-ness.
484
616
 
485
- A group's auto-skip re-arms exactly like a manual skip (same fingerprint
486
- mechanism) if the upstream inputs it depended on move. See
617
+ A manual/branch skip re-arms only after its upstream inputs move. An exclusive
618
+ group skip has an additional inverse: if the green winner is rejected or
619
+ retracted, a sibling whose latest skip reason is `exclusive` immediately
620
+ returns to `owed`; a structurally skipped sibling does not. See
487
621
  [`routing-groups.yaml`](../examples/workflows/routing-groups.yaml) for a
488
- runnable example, and [`docs/design.md` §26](design.md) for the full
489
- design (refusal timing, the judges interaction, and the model-checker parity
622
+ runnable example, and [`docs/design.md` §26](design.md) for the full design
623
+ (refusal timing, the judges interaction, and the model-checker parity
490
624
  guarantee).
491
625
 
492
626
  Eligibility (the automatic sweep) is pre-filtered the same way commit-time
@@ -560,7 +694,7 @@ steps:
560
694
  calls: delivery # bare local name; use <package>/<workflow> for a CAS target
561
695
  inputs: # child input → parent artifact (gate: sandbox green)
562
696
  proposal: proposal
563
- produces: [delivered] # one parent artifact; greens when delivery's output greens
697
+ produces: [delivered] # mirrors a green/value-defined output only after delivery is done
564
698
  - name: teardown
565
699
  consumes: [delivered]
566
700
  produces: [torn_down]
@@ -568,9 +702,12 @@ steps:
568
702
  body: Tear down and green `torn_down`.
569
703
  ```
570
704
 
571
- The engine spawns the child when the gate inputs are green, greens the parent's
572
- `calls:` output when the child's declared output greens (no Worker run), and re-provides
573
- inputs to the existing child if a gate input changes — it never spawns a duplicate.
705
+ The engine spawns the child when the gate inputs are green and keeps the parent's
706
+ `calls:` output owed until the child's declared output is green/value-defined **and**
707
+ `workflowDone(childDef, childArts)` says the whole child is done. Only then does it
708
+ mirror that value into the parent (no Worker run). A child with cleanup or a manually
709
+ supplied input still outstanding keeps the parent output owed. If a gate input changes,
710
+ the engine re-provides it to the existing child — it never spawns a duplicate.
574
711
 
575
712
  | | `include:` (Mode 1) | `calls:` (Mode 2) |
576
713
  |---|---|---|
package/docs/bundles.md CHANGED
@@ -161,6 +161,25 @@ This check is separate from the read-only file modes applied during install. Ins
161
161
 
162
162
  An executable file installed by an older Owenloop release may already have lost its execute bit because older hardening mapped every regular file to `0444`. Reinstalling the exact original `.wnlp` is a supported atomic repair for that same-digest object. Owenloop first applies strict archive parsing, manifest and runtime admission, configured signature and policy verification, per-file integrity verification, and canonical digest verification to the supplied archive. Owenloop then reconstructs and completely verifies a clean staged object before replacing the broken destination through the normal journaled directory swap. The broken object's bytes are never reused. The prior object remains as the swap backup until the unchanged same-digest index state is durably committed; a failed repair restores the recoverable prior object and index state. A successful repair preserves the existing coordinate-to-digest mapping, and a later reinstall follows normal idempotent deduplication.
163
163
 
164
+ ## Authenticated recovery for worker store misses
165
+
166
+ When a modern command or agent worker cannot resolve an exact order digest from
167
+ either local workflow store, it performs one authenticated recovery attempt.
168
+ Using that worker's scoped bearer, it reads raw canonical bytes from
169
+ `GET /api/bundles/<digest>`, required publication evidence from
170
+ `GET /api/publications/<digest>`, and optional origin evidence from
171
+ `GET /api/origins/<digest>`, then installs only into the global store. The
172
+ original store lookup is retried once after a successful install; a store hit
173
+ does no hub I/O.
174
+
175
+ Recovery is fail-closed. The installer compares the ingested canonical digest
176
+ with the order digest before any object or index mutation, and the required
177
+ pre-commit verifier validates the supplied publication/origin evidence under
178
+ the normal policy before the journaled install can commit. Missing or malformed
179
+ publication evidence, redirects, non-404 origin errors, body caps, timeouts,
180
+ content mismatches, invalid signatures, or policy refusals leave the store
181
+ unchanged and the worker reports the ordinary instruction-resolution failure.
182
+
164
183
  ## Bundle assets during execution
165
184
 
166
185
  When a command or agent step resolves its definition from an installed bundle,