@unbrained/pm-cli 2026.10.10 → 2026.10.11
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.
- package/.claude-plugin/marketplace.json +1 -36
- package/CHANGELOG.md +13 -4
- package/dist/cli-bundle/bundle-manifest.json +1 -382
- package/dist/cli-bundle/sdk-governance.js +1 -1
- package/dist/cli-bundle/sdk.js +9 -9
- package/dist/sdk/governance/artifact-ledgers.d.ts +59 -0
- package/dist/sdk/governance/artifact-ledgers.js +2 -0
- package/dist/sdk/governance.d.ts +1 -0
- package/dist/sdk/governance.js +1 -1
- package/dist/sdk/public-surface.json +1 -1
- package/docs/RELEASE_LEDGER_RECONCILIATION.md +148 -0
- package/docs/RELEASE_RELIABILITY.md +4 -0
- package/docs/RELEASING.md +34 -5
- package/docs/agent-task-token-baseline.json +1 -292
- package/docs/agent-task-transcripts.json +1 -478
- package/docs/ecosystem-review-2026-09-28.json +1 -2866
- package/docs/examples/policy-restricted-extension/manifest.json +1 -22
- package/docs/examples/policy-restricted-extension/package.json +1 -8
- package/docs/examples/sdk-agent-environment/package.json +1 -8
- package/docs/examples/sdk-app-embedding/package.json +1 -9
- package/docs/examples/sdk-contract-consumer/package.json +1 -11
- package/docs/examples/sdk-custom-tool/package.json +1 -13
- package/docs/examples/starter-extension/manifest.json +1 -33
- package/docs/examples/starter-extension/package.json +1 -10
- package/docs/performance/corpus-shape-comparison.json +1 -108
- package/docs/performance/latest-scale-report.json +1 -418
- package/docs/performance/scale-100k-report.json +1 -418
- package/docs/schemas/extension-manifest.schema.json +1 -165
- package/marketplace.json +1 -36
- package/package.json +1 -1
- package/packages/pm-beads/extensions/beads/manifest.json +1 -28
- package/packages/pm-beads/extensions/beads/package.json +1 -3
- package/packages/pm-beads/package.json +1 -50
- package/packages/pm-calendar/extensions/calendar/manifest.json +1 -22
- package/packages/pm-calendar/extensions/calendar/package.json +1 -3
- package/packages/pm-calendar/package.json +1 -51
- package/packages/pm-command-kit/extensions/command-kit/manifest.json +1 -22
- package/packages/pm-command-kit/extensions/command-kit/package.json +1 -3
- package/packages/pm-command-kit/package.json +1 -53
- package/packages/pm-digital-twin/extensions/digital-twin/manifest.json +1 -19
- package/packages/pm-digital-twin/extensions/digital-twin/package.json +1 -3
- package/packages/pm-digital-twin/package.json +1 -58
- package/packages/pm-governance-audit/extensions/governance-audit/manifest.json +1 -19
- package/packages/pm-governance-audit/extensions/governance-audit/package.json +1 -3
- package/packages/pm-governance-audit/package.json +1 -55
- package/packages/pm-guide-shell/extensions/guide-shell/manifest.json +1 -22
- package/packages/pm-guide-shell/extensions/guide-shell/package.json +1 -3
- package/packages/pm-guide-shell/package.json +1 -52
- package/packages/pm-kanban/extensions/kanban/manifest.json +1 -19
- package/packages/pm-kanban/extensions/kanban/package.json +1 -3
- package/packages/pm-kanban/package.json +1 -57
- package/packages/pm-lifecycle-hooks/extensions/lifecycle-hooks/manifest.json +1 -19
- package/packages/pm-lifecycle-hooks/extensions/lifecycle-hooks/package.json +1 -3
- package/packages/pm-lifecycle-hooks/package.json +1 -52
- package/packages/pm-linked-test-adapters/extensions/linked-test-adapters/manifest.json +1 -29
- package/packages/pm-linked-test-adapters/extensions/linked-test-adapters/package.json +1 -3
- package/packages/pm-linked-test-adapters/package.json +1 -53
- package/packages/pm-search-advanced/extensions/search-advanced/manifest.json +1 -29
- package/packages/pm-search-advanced/extensions/search-advanced/package.json +1 -3
- package/packages/pm-search-advanced/package.json +1 -54
- package/packages/pm-templates/extensions/templates/manifest.json +1 -22
- package/packages/pm-templates/extensions/templates/package.json +1 -3
- package/packages/pm-templates/package.json +1 -50
- package/packages/pm-todos/extensions/todos/manifest.json +1 -29
- package/packages/pm-todos/extensions/todos/package.json +1 -3
- package/packages/pm-todos/package.json +1 -51
- package/packages/pm-vcs/extensions/vcs/manifest.json +1 -19
- package/packages/pm-vcs/extensions/vcs/package.json +1 -3
- package/packages/pm-vcs/package.json +1 -55
- package/plugins/pm-claude/.claude-plugin/plugin.json +1 -25
- package/plugins/pm-claude/.mcp.json +1 -8
- package/plugins/pm-claude/hooks/hooks.json +1 -17
- package/plugins/pm-claude/package.json +1 -9
- package/plugins/pm-codex/.codex-plugin/plugin.json +1 -45
- package/plugins/pm-codex/.mcp.json +1 -11
- package/plugins/pm-codex/mcp.json +1 -11
- package/plugins/pm-codex/package.json +1 -9
- package/plugins/pm-codex/plugin.json +1 -45
- package/runtime-dependencies.json +1 -512
|
@@ -0,0 +1,148 @@
|
|
|
1
|
+
# Full-history release reconciliation
|
|
2
|
+
|
|
3
|
+
Owner: [pm-q91qyd](../.agents/pm/issues/pm-q91qyd.toon).
|
|
4
|
+
Upstream trigger evidence: [pm-prrlce](../.agents/pm/issues/pm-prrlce.toon).
|
|
5
|
+
|
|
6
|
+
The Release Reliability workflow compares every dated section in the
|
|
7
|
+
package-generated changelog, every release tag advertised by `origin`, and
|
|
8
|
+
every version in anonymous public npm metadata. It runs on the existing daily
|
|
9
|
+
schedule and completion triggers, even when a rolling reliability report fails.
|
|
10
|
+
Its separate `release-ledgers-<run>-<attempt>` artifact contains the complete
|
|
11
|
+
input inventories, counts, findings and exception dispositions for offline replay.
|
|
12
|
+
|
|
13
|
+
Missing public registry versions are declared-but-undelivered releases. Missing
|
|
14
|
+
changelog sections are delivered-but-undocumented releases. A tag missing its
|
|
15
|
+
section is tagged-but-unsectioned; an identity can belong to several classes.
|
|
16
|
+
Every missing ledger appears explicitly in the report, without counting one
|
|
17
|
+
identity several times or dropping overlapping classes.
|
|
18
|
+
The `classes` object names these three populations separately; `findings`
|
|
19
|
+
retains one row per discrepant identity.
|
|
20
|
+
|
|
21
|
+
The first complete census on 2026-10-10 found 14 discrepant identities among
|
|
22
|
+
143 sections, 151 tags and 143 public versions. The exact historical signatures
|
|
23
|
+
and their individual dispositions are recorded in
|
|
24
|
+
[the reviewed exception policy](../config/release-ledger-exceptions.json).
|
|
25
|
+
Unpublished historical preparations are retained as abandoned publications;
|
|
26
|
+
historical documentation gaps remain visible without inventing shipped content.
|
|
27
|
+
These are explicit exceptions, not claims that the three historical sets agree.
|
|
28
|
+
No tag is deleted and no old artifact is published by this reporting path.
|
|
29
|
+
|
|
30
|
+
An exception matches one identity and one exact missing-ledger signature.
|
|
31
|
+
Additional missing evidence invalidates it. Correcting a historical gap makes
|
|
32
|
+
its exception stale and requires removing or revising that policy entry in a
|
|
33
|
+
reviewed change. A new discrepancy or stale exception fails the report.
|
|
34
|
+
Empty total inventories, incomplete populations, duplicate identities, malformed
|
|
35
|
+
remote output, registry failures and missing exception evidence refuse evaluation.
|
|
36
|
+
|
|
37
|
+
The registry request uses the same public npm metadata surface as release
|
|
38
|
+
acceptance, with explicit public registry selection, empty user/global
|
|
39
|
+
configuration and a fresh temporary cache. It runs outside the checkout and
|
|
40
|
+
removes inherited npm authentication/configuration. The report never archives
|
|
41
|
+
credentials, command output or environment values.
|
|
42
|
+
On Windows the collector invokes `cmd.exe` explicitly for npm's `.cmd` launcher.
|
|
43
|
+
The package name must satisfy the restricted registry identity grammar before
|
|
44
|
+
any child starts; the remaining arguments are fixed. Required Windows acceptance
|
|
45
|
+
executes real Git and public npm requests from paths containing spaces and `&`,
|
|
46
|
+
first proves that direct `npm.cmd` execution fails, then verifies the complete
|
|
47
|
+
inventory and temporary-directory cleanup through the repaired launcher.
|
|
48
|
+
No `shell: true` argument-array fallback or resource ceiling increase is used.
|
|
49
|
+
The parser supports npm 11's object and npm 12's single-result array, with
|
|
50
|
+
[captured public npm 12 metadata](../tests/fixtures/release-ledgers/npm-view-12.json)
|
|
51
|
+
as its format regression. Multiple package results refuse evaluation.
|
|
52
|
+
|
|
53
|
+
To run the actual collector after building:
|
|
54
|
+
|
|
55
|
+
```bash
|
|
56
|
+
RELEASE_LEDGERS_OUTPUT=/absolute/report.json node --input-type=module -e '
|
|
57
|
+
import { main } from "./scripts/release/collect-release-ledgers.mjs";
|
|
58
|
+
const report = main();
|
|
59
|
+
console.log(JSON.stringify({ok:report.ok,counts:report.counts,findings:report.findings}));
|
|
60
|
+
process.exitCode = report.ok ? 0 : 1;
|
|
61
|
+
'
|
|
62
|
+
```
|
|
63
|
+
|
|
64
|
+
Packages and non-release projects can use the same pure SDK primitive with
|
|
65
|
+
their own named artifact inventories and decision references:
|
|
66
|
+
|
|
67
|
+
```ts
|
|
68
|
+
import { reconcileArtifactLedgers } from "@unbrained/pm-cli/sdk/governance";
|
|
69
|
+
|
|
70
|
+
const report = reconcileArtifactLedgers([
|
|
71
|
+
{ name: "declared", complete: true, identities: ["artifact-1"] },
|
|
72
|
+
{ name: "delivered", complete: true, identities: ["artifact-1"] },
|
|
73
|
+
]);
|
|
74
|
+
```
|
|
75
|
+
|
|
76
|
+
`ArtifactLedgerSnapshot`, `ArtifactLedgerException`, `ArtifactLedgerFinding`
|
|
77
|
+
and `ArtifactLedgerReport` describe this harness-independent contract.
|
|
78
|
+
The evaluator performs no filesystem, clock or network operations and preserves
|
|
79
|
+
caller inputs. For offline replay, feed the saved `snapshots` and `exceptions`
|
|
80
|
+
back into `reconcileArtifactLedgers`.
|
|
81
|
+
|
|
82
|
+
## Tag-carried trigger declarations
|
|
83
|
+
|
|
84
|
+
New preparations create annotated tags whose bounded JSON message binds the
|
|
85
|
+
declared upstream event/origin, repository, run/attempt, reviewed source commit
|
|
86
|
+
and prepared target commit. Before publication, the Release workflow checks
|
|
87
|
+
the tag's peeled commit and source ancestry and compares the source declaration
|
|
88
|
+
with the exact GitHub run attempt, including repository, event, source SHA,
|
|
89
|
+
the producer workflow path and declared Auto Release run name. It independently
|
|
90
|
+
fetches the repository's default branch and requires the source run's branch to
|
|
91
|
+
match it. Auto Release uploads an immutable producer binding before pushing;
|
|
92
|
+
publication independently downloads that run attempt's artifact and compares
|
|
93
|
+
the exact tag name, target commit, tag object and canonical declaration. Missing,
|
|
94
|
+
expired, ambiguous or mismatched evidence stops publication; the artifact is
|
|
95
|
+
retained for 90 days. The verifier never assumes the default branch is named
|
|
96
|
+
`main`.
|
|
97
|
+
|
|
98
|
+
The separate `release-publication-origin-<run>-<attempt>` artifact retains the
|
|
99
|
+
tag-object SHA, upstream declaration, independent verification result and the
|
|
100
|
+
current publication/recovery trigger. An operator recovering a dispatcher
|
|
101
|
+
release therefore retains both origins. A rejected atomic push after sealing
|
|
102
|
+
refuses to rebase into a different target; a new preparation must establish a
|
|
103
|
+
new independent binding. Local preparation remains supported without pushing;
|
|
104
|
+
local `--push` refuses before Git inspection or release-asset mutation.
|
|
105
|
+
|
|
106
|
+
Historical lightweight tags and unrelated legacy annotations stay explicitly
|
|
107
|
+
unattributed only when publication matches the repository, tag name and exact
|
|
108
|
+
content-addressed object in the
|
|
109
|
+
[frozen pre-rollout inventory](../config/release-tag-legacy.json). That reviewed
|
|
110
|
+
inventory identifies all 152 origin tag objects observed before this rollout;
|
|
111
|
+
it is not extended automatically. Unknown lightweight tags, unrelated new
|
|
112
|
+
annotations and rewritten historical objects stop before publication, including
|
|
113
|
+
operator recovery. A missing or unreadable inventory fails closed. No origin is
|
|
114
|
+
inferred from a tag-push event or actor. Local preparation records an operator
|
|
115
|
+
declaration but hosted publication refuses it without a hosted producer.
|
|
116
|
+
Tag objects are content-addressed evidence; this does not assert that GitHub
|
|
117
|
+
release/tag immutability policy has been enabled.
|
|
118
|
+
|
|
119
|
+
The downstream tag-push workflow's run name still names `tag_push`; its verified
|
|
120
|
+
upstream origin lives in the summary and artifact. The remaining run-name and
|
|
121
|
+
operator-only manual follow-up contracts retain their existing owners
|
|
122
|
+
[pm-prrlce](../.agents/pm/issues/pm-prrlce.toon) and
|
|
123
|
+
[pm-uenmu9](../.agents/pm/issues/pm-uenmu9.toon). The one-automatic-release-per-UTC-day
|
|
124
|
+
guard and exact-tag recovery semantics are unchanged.
|
|
125
|
+
|
|
126
|
+
## Installed acceptance path cost
|
|
127
|
+
|
|
128
|
+
Owner: [pm-release-acceptance-path-cost](../.agents/pm/issues/pm-release-acceptance-path-cost.toon).
|
|
129
|
+
|
|
130
|
+
The disposable installed-consumer layout uses concise directory names because
|
|
131
|
+
`init` repeats its workspace path throughout its complete JSON guidance. Long
|
|
132
|
+
macOS temporary ancestry previously made an otherwise valid public artifact
|
|
133
|
+
exceed the raw 12,000-character init ceiling. Shorter private paths retain every
|
|
134
|
+
output field, lifecycle assertion, deadline and ceiling. The
|
|
135
|
+
[captured public output regression](../tests/fixtures/release-acceptance/README.md)
|
|
136
|
+
checks the raw cost for independent control/candidate local npm, global npm and
|
|
137
|
+
Bun layouts; actual installed sessions remain separate acceptance evidence.
|
|
138
|
+
|
|
139
|
+
## Publication metadata weight
|
|
140
|
+
|
|
141
|
+
Owner: [pm-998juj](../.agents/pm/tasks/pm-998juj.toon).
|
|
142
|
+
|
|
143
|
+
The publication stage compacts first-party JSON while leaving checkout files
|
|
144
|
+
readable and third-party runtime files unchanged. Every parsed value must survive
|
|
145
|
+
serialization exactly; invalid JSON and lossy representations, including negative
|
|
146
|
+
zero or overflowing numbers, stop packing. Real tarball regressions verify nested
|
|
147
|
+
extension data, Unicode, escaped text and the runtime ledger. Both committed size
|
|
148
|
+
profiles, required files and source-map exclusions remain enforced.
|
|
@@ -34,6 +34,10 @@ by release preparation, not the inferred cron occurrence. A delayed run cannot
|
|
|
34
34
|
backdate a tag. Reporting never creates, retargets, retries, or publishes a tag.
|
|
35
35
|
The unchanged [release recovery rules](RELEASING.md#failure-handling) apply.
|
|
36
36
|
|
|
37
|
+
The independent [full-history ledger report](RELEASE_LEDGER_RECONCILIATION.md)
|
|
38
|
+
also reconciles public registry versions, release tags and generated sections.
|
|
39
|
+
It preserves every historical exception and fails on new or changed drift.
|
|
40
|
+
|
|
37
41
|
## Evidence and denominator
|
|
38
42
|
|
|
39
43
|
The collector paginates Auto Release runs over an explicit half-open UTC window
|
package/docs/RELEASING.md
CHANGED
|
@@ -172,6 +172,17 @@ Pipeline entrypoint:
|
|
|
172
172
|
node scripts/release/run-release-pipeline.mjs
|
|
173
173
|
```
|
|
174
174
|
|
|
175
|
+
Tracked by [pm-prrlce](../.agents/pm/issues/pm-prrlce.toon): hosted preparation
|
|
176
|
+
uses `--push --defer-push` to write an exact producer binding. Auto Release uploads
|
|
177
|
+
`release-producer-binding-<attempt>` before verifying and atomically pushing its
|
|
178
|
+
prepared refs. Publication independently fetches the verified run's artifact and
|
|
179
|
+
compares the tag name, target commit, tag object and declaration. Missing, expired,
|
|
180
|
+
ambiguous or mismatched evidence fails closed; artifact retention is 90 days.
|
|
181
|
+
A branch advance after sealing refuses publication rather than rebasing into a
|
|
182
|
+
different target. Local `--push` refuses before Git inspection or asset mutation;
|
|
183
|
+
local preparation without pushing remains supported. Historical admission remains
|
|
184
|
+
limited to the reviewed exact-object inventory.
|
|
185
|
+
|
|
175
186
|
The pipeline performs:
|
|
176
187
|
|
|
177
188
|
1. change detection + one-release-per-day guard
|
|
@@ -421,14 +432,28 @@ maintainer supplies
|
|
|
421
432
|
pair that acknowledgement with migration guidance, compatible extension
|
|
422
433
|
version bounds where applicable, and the next eligible date-based release.
|
|
423
434
|
|
|
424
|
-
5.
|
|
435
|
+
5. Land the implementation through a reviewed pull request, then use hosted
|
|
436
|
+
preparation for publication.
|
|
437
|
+
|
|
438
|
+
Local preparation without `--push` is a diagnostic preview. Its local producer
|
|
439
|
+
declaration cannot authorize publication. Once the reviewed implementation is on
|
|
440
|
+
`main`, dispatch the existing Auto Release workflow from that default branch:
|
|
425
441
|
|
|
426
442
|
```bash
|
|
427
|
-
|
|
428
|
-
|
|
429
|
-
git push origin v<version>
|
|
443
|
+
gh workflow run auto-release.yml --ref main \
|
|
444
|
+
-f push=true -f dry_run=false -f telemetry_mode=off -f trigger_origin=operator
|
|
430
445
|
```
|
|
431
446
|
|
|
447
|
+
Hosted preparation runs the mandatory gates, creates the version commit and
|
|
448
|
+
annotated tag, and seals their exact producer binding in an immutable artifact
|
|
449
|
+
before verifying and atomically pushing either ref. It waits for the single
|
|
450
|
+
tag-driven Release workflow to publish and accept that exact package. Do not push
|
|
451
|
+
locally prepared tags or invoke `npm publish` separately.
|
|
452
|
+
|
|
453
|
+
If today's immutable tag already exists, use the exact-tag recovery procedure
|
|
454
|
+
below. A local diagnostic tag does not prove hosted preparation or publication;
|
|
455
|
+
never move a published tag or mint a second automatic version for the same UTC day.
|
|
456
|
+
|
|
432
457
|
## GitHub Workflow
|
|
433
458
|
|
|
434
459
|
`.github/workflows/release.yml` runs on `v*.*.*` tags and handles:
|
|
@@ -454,7 +479,11 @@ git push origin v<version>
|
|
|
454
479
|
- a base npm pack dry run, optional Sentry debug-ID injection, a second
|
|
455
480
|
`sentry-injected` packlist budget over the exact publishable bytes, and only
|
|
456
481
|
then optional Sentry release metadata/upload/finalization and the npx tarball
|
|
457
|
-
smoke test
|
|
482
|
+
smoke test. Sentry release metadata and sourcemaps are uploaded only from
|
|
483
|
+
tag sources. Recovery of an already-published version from reviewed `main`
|
|
484
|
+
preserves that version's existing Sentry commits and sourcemaps; local debug-ID
|
|
485
|
+
injection still validates the candidate artifact budget without changing the
|
|
486
|
+
published release's symbolication data.
|
|
458
487
|
- generated release notes from changelog plus sanitized tracker metadata. The
|
|
459
488
|
tracker read projects only release-relevant fields and explicitly removes row
|
|
460
489
|
and token limits. The generator validates the completeness receipt before
|
|
@@ -1,292 +1 @@
|
|
|
1
|
-
{
|
|
2
|
-
"version": 5,
|
|
3
|
-
"transcript_version": 2,
|
|
4
|
-
"transcript_digest": "sha256:6d4041acfaa4075dde1972d6127a74f18916f7661738b4f5d5fd47a61c179a78",
|
|
5
|
-
"estimator": "ceil(utf8_bytes / 4)",
|
|
6
|
-
"comparison_scope": "plan_recovery_tracker_root_canonicalized",
|
|
7
|
-
"measurement_scope": "output_before_token_accounting",
|
|
8
|
-
"published_with_release": true,
|
|
9
|
-
"orientation": {
|
|
10
|
-
"canonical_task_id": "orientation-context-intent",
|
|
11
|
-
"required_capabilities": [
|
|
12
|
-
"actionable_work",
|
|
13
|
-
"active_ownership",
|
|
14
|
-
"workspace_state"
|
|
15
|
-
],
|
|
16
|
-
"measured_winner_tokens": 1087,
|
|
17
|
-
"protocols": [
|
|
18
|
-
{
|
|
19
|
-
"task_id": "orientation-context-intent",
|
|
20
|
-
"command_count": 1,
|
|
21
|
-
"estimated_tokens": 1087,
|
|
22
|
-
"capabilities": [
|
|
23
|
-
"actionable_work",
|
|
24
|
-
"active_ownership",
|
|
25
|
-
"workspace_state"
|
|
26
|
-
],
|
|
27
|
-
"capability_evidence": {
|
|
28
|
-
"actionable_work": [
|
|
29
|
-
{
|
|
30
|
-
"step_id": "context",
|
|
31
|
-
"field_path": "hierarchy"
|
|
32
|
-
}
|
|
33
|
-
],
|
|
34
|
-
"active_ownership": [
|
|
35
|
-
{
|
|
36
|
-
"step_id": "context",
|
|
37
|
-
"field_path": "activity"
|
|
38
|
-
}
|
|
39
|
-
],
|
|
40
|
-
"workspace_state": [
|
|
41
|
-
{
|
|
42
|
-
"step_id": "context",
|
|
43
|
-
"field_path": "summary.active_items"
|
|
44
|
-
}
|
|
45
|
-
]
|
|
46
|
-
}
|
|
47
|
-
},
|
|
48
|
-
{
|
|
49
|
-
"task_id": "orientation-operating-rules",
|
|
50
|
-
"command_count": 4,
|
|
51
|
-
"estimated_tokens": 2325,
|
|
52
|
-
"capabilities": [
|
|
53
|
-
"actionable_work",
|
|
54
|
-
"active_ownership",
|
|
55
|
-
"workspace_state"
|
|
56
|
-
],
|
|
57
|
-
"capability_evidence": {
|
|
58
|
-
"actionable_work": [
|
|
59
|
-
{
|
|
60
|
-
"step_id": "open",
|
|
61
|
-
"field_path": "items"
|
|
62
|
-
}
|
|
63
|
-
],
|
|
64
|
-
"active_ownership": [
|
|
65
|
-
{
|
|
66
|
-
"step_id": "in-progress",
|
|
67
|
-
"field_path": "items"
|
|
68
|
-
}
|
|
69
|
-
],
|
|
70
|
-
"workspace_state": [
|
|
71
|
-
{
|
|
72
|
-
"step_id": "context",
|
|
73
|
-
"field_path": "summary.active_items"
|
|
74
|
-
}
|
|
75
|
-
]
|
|
76
|
-
}
|
|
77
|
-
},
|
|
78
|
-
{
|
|
79
|
-
"task_id": "orientation-contracts-next",
|
|
80
|
-
"command_count": 2,
|
|
81
|
-
"estimated_tokens": 5203,
|
|
82
|
-
"capabilities": [
|
|
83
|
-
"actionable_work",
|
|
84
|
-
"active_ownership",
|
|
85
|
-
"workspace_state"
|
|
86
|
-
],
|
|
87
|
-
"capability_evidence": {
|
|
88
|
-
"actionable_work": [
|
|
89
|
-
{
|
|
90
|
-
"step_id": "next",
|
|
91
|
-
"field_path": "recommended.id"
|
|
92
|
-
}
|
|
93
|
-
],
|
|
94
|
-
"active_ownership": [
|
|
95
|
-
{
|
|
96
|
-
"step_id": "next",
|
|
97
|
-
"field_path": "held_by_others"
|
|
98
|
-
}
|
|
99
|
-
],
|
|
100
|
-
"workspace_state": [
|
|
101
|
-
{
|
|
102
|
-
"step_id": "next",
|
|
103
|
-
"field_path": "summary.in_progress"
|
|
104
|
-
}
|
|
105
|
-
]
|
|
106
|
-
}
|
|
107
|
-
}
|
|
108
|
-
]
|
|
109
|
-
},
|
|
110
|
-
"tasks": [
|
|
111
|
-
{
|
|
112
|
-
"id": "orientation-context-intent",
|
|
113
|
-
"max_estimated_tokens": 1087,
|
|
114
|
-
"steps": [
|
|
115
|
-
{
|
|
116
|
-
"id": "context",
|
|
117
|
-
"max_estimated_tokens": 1087,
|
|
118
|
-
"accounting_mode": "self_reported"
|
|
119
|
-
}
|
|
120
|
-
]
|
|
121
|
-
},
|
|
122
|
-
{
|
|
123
|
-
"id": "orientation-contracts-next",
|
|
124
|
-
"max_estimated_tokens": 5203,
|
|
125
|
-
"steps": [
|
|
126
|
-
{
|
|
127
|
-
"id": "contracts",
|
|
128
|
-
"max_estimated_tokens": 4481,
|
|
129
|
-
"accounting_mode": "self_reported"
|
|
130
|
-
},
|
|
131
|
-
{
|
|
132
|
-
"id": "next",
|
|
133
|
-
"max_estimated_tokens": 722,
|
|
134
|
-
"accounting_mode": "self_reported"
|
|
135
|
-
}
|
|
136
|
-
]
|
|
137
|
-
},
|
|
138
|
-
{
|
|
139
|
-
"id": "orientation-operating-rules",
|
|
140
|
-
"max_estimated_tokens": 2325,
|
|
141
|
-
"steps": [
|
|
142
|
-
{
|
|
143
|
-
"id": "context",
|
|
144
|
-
"max_estimated_tokens": 1087,
|
|
145
|
-
"accounting_mode": "self_reported"
|
|
146
|
-
},
|
|
147
|
-
{
|
|
148
|
-
"id": "search",
|
|
149
|
-
"max_estimated_tokens": 164,
|
|
150
|
-
"accounting_mode": "self_reported"
|
|
151
|
-
},
|
|
152
|
-
{
|
|
153
|
-
"id": "open",
|
|
154
|
-
"max_estimated_tokens": 878,
|
|
155
|
-
"accounting_mode": "self_reported"
|
|
156
|
-
},
|
|
157
|
-
{
|
|
158
|
-
"id": "in-progress",
|
|
159
|
-
"max_estimated_tokens": 196,
|
|
160
|
-
"accounting_mode": "self_reported"
|
|
161
|
-
}
|
|
162
|
-
]
|
|
163
|
-
},
|
|
164
|
-
{
|
|
165
|
-
"id": "context-bootstrap",
|
|
166
|
-
"max_estimated_tokens": 2010,
|
|
167
|
-
"steps": [
|
|
168
|
-
{
|
|
169
|
-
"id": "triage",
|
|
170
|
-
"max_estimated_tokens": 552,
|
|
171
|
-
"accounting_mode": "self_reported"
|
|
172
|
-
},
|
|
173
|
-
{
|
|
174
|
-
"id": "orient",
|
|
175
|
-
"max_estimated_tokens": 1087,
|
|
176
|
-
"accounting_mode": "self_reported"
|
|
177
|
-
},
|
|
178
|
-
{
|
|
179
|
-
"id": "inspect",
|
|
180
|
-
"max_estimated_tokens": 371,
|
|
181
|
-
"accounting_mode": "self_reported"
|
|
182
|
-
}
|
|
183
|
-
]
|
|
184
|
-
},
|
|
185
|
-
{
|
|
186
|
-
"id": "closed-domain-recovery",
|
|
187
|
-
"max_estimated_tokens": 630,
|
|
188
|
-
"steps": [
|
|
189
|
-
{
|
|
190
|
-
"id": "refuse-intent",
|
|
191
|
-
"max_estimated_tokens": 259,
|
|
192
|
-
"accounting_mode": "self_reported"
|
|
193
|
-
},
|
|
194
|
-
{
|
|
195
|
-
"id": "retry-intent",
|
|
196
|
-
"max_estimated_tokens": 371,
|
|
197
|
-
"accounting_mode": "self_reported"
|
|
198
|
-
}
|
|
199
|
-
]
|
|
200
|
-
},
|
|
201
|
-
{
|
|
202
|
-
"id": "unknown-option-recovery",
|
|
203
|
-
"max_estimated_tokens": 596,
|
|
204
|
-
"steps": [
|
|
205
|
-
{
|
|
206
|
-
"id": "refuse-option",
|
|
207
|
-
"max_estimated_tokens": 433,
|
|
208
|
-
"accounting_mode": "self_reported"
|
|
209
|
-
},
|
|
210
|
-
{
|
|
211
|
-
"id": "retry-option",
|
|
212
|
-
"max_estimated_tokens": 163,
|
|
213
|
-
"accounting_mode": "self_reported"
|
|
214
|
-
}
|
|
215
|
-
]
|
|
216
|
-
},
|
|
217
|
-
{
|
|
218
|
-
"id": "lifecycle-mutation",
|
|
219
|
-
"max_estimated_tokens": 812,
|
|
220
|
-
"steps": [
|
|
221
|
-
{
|
|
222
|
-
"id": "create",
|
|
223
|
-
"max_estimated_tokens": 20,
|
|
224
|
-
"accounting_mode": "self_reported"
|
|
225
|
-
},
|
|
226
|
-
{
|
|
227
|
-
"id": "inspect-open",
|
|
228
|
-
"max_estimated_tokens": 363,
|
|
229
|
-
"accounting_mode": "self_reported"
|
|
230
|
-
},
|
|
231
|
-
{
|
|
232
|
-
"id": "close",
|
|
233
|
-
"max_estimated_tokens": 31,
|
|
234
|
-
"accounting_mode": "self_reported"
|
|
235
|
-
},
|
|
236
|
-
{
|
|
237
|
-
"id": "inspect-closed",
|
|
238
|
-
"max_estimated_tokens": 398,
|
|
239
|
-
"accounting_mode": "self_reported"
|
|
240
|
-
}
|
|
241
|
-
]
|
|
242
|
-
},
|
|
243
|
-
{
|
|
244
|
-
"id": "bulk-effect-outcomes",
|
|
245
|
-
"max_estimated_tokens": 326,
|
|
246
|
-
"steps": [
|
|
247
|
-
{
|
|
248
|
-
"id": "create-bulk-target",
|
|
249
|
-
"max_estimated_tokens": 20,
|
|
250
|
-
"accounting_mode": "self_reported"
|
|
251
|
-
},
|
|
252
|
-
{
|
|
253
|
-
"id": "partial-effect",
|
|
254
|
-
"max_estimated_tokens": 174,
|
|
255
|
-
"accounting_mode": "self_reported"
|
|
256
|
-
},
|
|
257
|
-
{
|
|
258
|
-
"id": "no-effect",
|
|
259
|
-
"max_estimated_tokens": 132,
|
|
260
|
-
"accounting_mode": "self_reported"
|
|
261
|
-
}
|
|
262
|
-
]
|
|
263
|
-
},
|
|
264
|
-
{
|
|
265
|
-
"id": "plan-execution-mutations",
|
|
266
|
-
"max_estimated_tokens": 806,
|
|
267
|
-
"steps": [
|
|
268
|
-
{
|
|
269
|
-
"id": "create-plan",
|
|
270
|
-
"max_estimated_tokens": 21,
|
|
271
|
-
"accounting_mode": "self_reported"
|
|
272
|
-
},
|
|
273
|
-
{
|
|
274
|
-
"id": "add-step",
|
|
275
|
-
"max_estimated_tokens": 254,
|
|
276
|
-
"accounting_mode": "self_reported"
|
|
277
|
-
},
|
|
278
|
-
{
|
|
279
|
-
"id": "complete-step",
|
|
280
|
-
"max_estimated_tokens": 253,
|
|
281
|
-
"accounting_mode": "self_reported"
|
|
282
|
-
},
|
|
283
|
-
{
|
|
284
|
-
"id": "inspect-plan",
|
|
285
|
-
"max_estimated_tokens": 278,
|
|
286
|
-
"accounting_mode": "self_reported"
|
|
287
|
-
}
|
|
288
|
-
]
|
|
289
|
-
}
|
|
290
|
-
],
|
|
291
|
-
"composite_max_estimated_tokens": 13795
|
|
292
|
-
}
|
|
1
|
+
{"version":5,"transcript_version":2,"transcript_digest":"sha256:6d4041acfaa4075dde1972d6127a74f18916f7661738b4f5d5fd47a61c179a78","estimator":"ceil(utf8_bytes / 4)","comparison_scope":"plan_recovery_tracker_root_canonicalized","measurement_scope":"output_before_token_accounting","published_with_release":true,"orientation":{"canonical_task_id":"orientation-context-intent","required_capabilities":["actionable_work","active_ownership","workspace_state"],"measured_winner_tokens":1087,"protocols":[{"task_id":"orientation-context-intent","command_count":1,"estimated_tokens":1087,"capabilities":["actionable_work","active_ownership","workspace_state"],"capability_evidence":{"actionable_work":[{"step_id":"context","field_path":"hierarchy"}],"active_ownership":[{"step_id":"context","field_path":"activity"}],"workspace_state":[{"step_id":"context","field_path":"summary.active_items"}]}},{"task_id":"orientation-operating-rules","command_count":4,"estimated_tokens":2325,"capabilities":["actionable_work","active_ownership","workspace_state"],"capability_evidence":{"actionable_work":[{"step_id":"open","field_path":"items"}],"active_ownership":[{"step_id":"in-progress","field_path":"items"}],"workspace_state":[{"step_id":"context","field_path":"summary.active_items"}]}},{"task_id":"orientation-contracts-next","command_count":2,"estimated_tokens":5203,"capabilities":["actionable_work","active_ownership","workspace_state"],"capability_evidence":{"actionable_work":[{"step_id":"next","field_path":"recommended.id"}],"active_ownership":[{"step_id":"next","field_path":"held_by_others"}],"workspace_state":[{"step_id":"next","field_path":"summary.in_progress"}]}}]},"tasks":[{"id":"orientation-context-intent","max_estimated_tokens":1087,"steps":[{"id":"context","max_estimated_tokens":1087,"accounting_mode":"self_reported"}]},{"id":"orientation-contracts-next","max_estimated_tokens":5203,"steps":[{"id":"contracts","max_estimated_tokens":4481,"accounting_mode":"self_reported"},{"id":"next","max_estimated_tokens":722,"accounting_mode":"self_reported"}]},{"id":"orientation-operating-rules","max_estimated_tokens":2325,"steps":[{"id":"context","max_estimated_tokens":1087,"accounting_mode":"self_reported"},{"id":"search","max_estimated_tokens":164,"accounting_mode":"self_reported"},{"id":"open","max_estimated_tokens":878,"accounting_mode":"self_reported"},{"id":"in-progress","max_estimated_tokens":196,"accounting_mode":"self_reported"}]},{"id":"context-bootstrap","max_estimated_tokens":2010,"steps":[{"id":"triage","max_estimated_tokens":552,"accounting_mode":"self_reported"},{"id":"orient","max_estimated_tokens":1087,"accounting_mode":"self_reported"},{"id":"inspect","max_estimated_tokens":371,"accounting_mode":"self_reported"}]},{"id":"closed-domain-recovery","max_estimated_tokens":630,"steps":[{"id":"refuse-intent","max_estimated_tokens":259,"accounting_mode":"self_reported"},{"id":"retry-intent","max_estimated_tokens":371,"accounting_mode":"self_reported"}]},{"id":"unknown-option-recovery","max_estimated_tokens":596,"steps":[{"id":"refuse-option","max_estimated_tokens":433,"accounting_mode":"self_reported"},{"id":"retry-option","max_estimated_tokens":163,"accounting_mode":"self_reported"}]},{"id":"lifecycle-mutation","max_estimated_tokens":812,"steps":[{"id":"create","max_estimated_tokens":20,"accounting_mode":"self_reported"},{"id":"inspect-open","max_estimated_tokens":363,"accounting_mode":"self_reported"},{"id":"close","max_estimated_tokens":31,"accounting_mode":"self_reported"},{"id":"inspect-closed","max_estimated_tokens":398,"accounting_mode":"self_reported"}]},{"id":"bulk-effect-outcomes","max_estimated_tokens":326,"steps":[{"id":"create-bulk-target","max_estimated_tokens":20,"accounting_mode":"self_reported"},{"id":"partial-effect","max_estimated_tokens":174,"accounting_mode":"self_reported"},{"id":"no-effect","max_estimated_tokens":132,"accounting_mode":"self_reported"}]},{"id":"plan-execution-mutations","max_estimated_tokens":806,"steps":[{"id":"create-plan","max_estimated_tokens":21,"accounting_mode":"self_reported"},{"id":"add-step","max_estimated_tokens":254,"accounting_mode":"self_reported"},{"id":"complete-step","max_estimated_tokens":253,"accounting_mode":"self_reported"},{"id":"inspect-plan","max_estimated_tokens":278,"accounting_mode":"self_reported"}]}],"composite_max_estimated_tokens":13795}
|