@skitterbyte/skitterspec-linear 19.3.0 → 19.5.0

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 (71) hide show
  1. package/assets/claude-md-section.md +20 -1
  2. package/assets/core/SETUP.md +5 -0
  3. package/assets/core/env.config.json.example +1 -0
  4. package/assets/core/env.config.md +143 -8
  5. package/assets/core/linear.config.json.example +2 -2
  6. package/assets/core/linear.config.md +49 -16
  7. package/assets/hooks/main-guard.cjs +40 -6
  8. package/assets/hooks/review-gate.cjs +42 -7
  9. package/assets/rules/commit-trailers.md +24 -1
  10. package/assets/rules/spec-planning.md +81 -18
  11. package/assets/rules/spec-reports.md +1 -1
  12. package/assets/skills/spec/SKILL.md +55 -2
  13. package/assets/skills/spec-bug/SKILL.md +36 -3
  14. package/assets/skills/spec-cancel/SKILL.md +37 -0
  15. package/assets/skills/spec-claim/SKILL.md +41 -2
  16. package/assets/skills/spec-complete/SKILL.md +115 -0
  17. package/assets/skills/spec-hotfix/SKILL.md +51 -4
  18. package/assets/skills/spec-incident/SKILL.md +22 -0
  19. package/assets/skills/spec-linear-setup/SKILL.md +23 -1
  20. package/assets/skills/spec-next/SKILL.md +12 -5
  21. package/assets/skills/spec-push/SKILL.md +22 -0
  22. package/assets/skills/spec-review/SKILL.md +34 -0
  23. package/assets/skills/spec-start/SKILL.md +50 -3
  24. package/assets/skills/spec-status/SKILL.md +18 -3
  25. package/assets/skills/spec-sync/SKILL.md +1 -0
  26. package/assets/skills/spec-to-main/SKILL.md +47 -1
  27. package/package.json +1 -1
  28. package/src/cli.js +1188 -79
  29. package/src/env/base.js +104 -0
  30. package/src/env/bootstrap.js +22 -3
  31. package/src/env/classify.js +12 -4
  32. package/src/env/config.js +77 -5
  33. package/src/env/disposition.js +62 -0
  34. package/src/env/handoff.js +81 -0
  35. package/src/env/host.js +262 -0
  36. package/src/env/hotfix.js +237 -9
  37. package/src/env/hotfixland.js +105 -10
  38. package/src/env/inflight.js +98 -0
  39. package/src/env/integrate.js +136 -9
  40. package/src/env/live.js +8 -7
  41. package/src/env/mainguard.js +20 -5
  42. package/src/env/phases.js +28 -0
  43. package/src/env/project.js +223 -0
  44. package/src/env/provision.js +30 -8
  45. package/src/env/resolve.js +147 -25
  46. package/src/env/review.js +21 -7
  47. package/src/env/reviewers.js +7 -10
  48. package/src/env/serve.js +27 -11
  49. package/src/env/specstate.js +101 -0
  50. package/src/env/supervise.js +15 -2
  51. package/src/env/teardown.js +46 -3
  52. package/src/init.js +49 -9
  53. package/src/vendor/linear/cli-sanitise.js +8 -3
  54. package/src/vendor/linear/cli-sync.js +64 -10
  55. package/src/vendor/linear/config.js +27 -21
  56. package/src/vendor/linear/doctor.js +2 -0
  57. package/src/vendor/provider-kit/index.js +4 -0
  58. package/src/vendor/provider-kit/src/apply.js +58 -6
  59. package/src/vendor/provider-kit/src/assignment.js +78 -1
  60. package/src/vendor/provider-kit/src/config.js +34 -1
  61. package/src/vendor/provider-kit/src/credentials.js +7 -2
  62. package/src/vendor/provider-kit/src/list-rows.js +35 -11
  63. package/src/vendor/provider-kit/src/released.js +48 -4
  64. package/src/vendor/provider-kit/src/specs.js +18 -7
  65. package/src/vendor/provider-kit/src/splice.js +64 -0
  66. package/src/vendor/provider-kit/src/status.js +137 -0
  67. package/src/vendor/sync-core/index.js +14 -1
  68. package/src/vendor/sync-core/src/compare.js +81 -6
  69. package/src/vendor/sync-core/src/normalize.js +50 -7
  70. package/src/vendor/sync-core/src/push.js +65 -5
  71. package/src/vendor/sync-core/src/summary.js +257 -0
@@ -7,7 +7,7 @@ and lifecycle stay consistent. Each sets a status on the spec header
7
7
 
8
8
  | Skill | Purpose | Status | Folder |
9
9
  |-------|---------|--------|--------|
10
- | `/spec` | (Feature/Chore/Issue) Grill to a clear shared understanding, propose the kind, write a groomed spec in its own worktree, then land it on the verdict | `Ready` (or `Draft`) | `specs/backlog/` (on the base branch, by a fast-forward) |
10
+ | `/spec` | (Feature/Chore/Issue) Grill to a clear shared understanding, propose the kind, write a groomed spec in its own worktree, then land it on the verdict | `Ready` (or `Draft`) | `specs/backlog/` (on the base branch by a fast-forward — or riding the feature branch under `land.specDocs: "branch"`) |
11
11
  | `/spec-bug` | (Bug) Reproduce with a failing test, capture spec, drive red→green | `In Progress` | `specs/in-progress/` |
12
12
  | `/spec-hotfix` | (Hotfix) Fork a worktree from a release tag, red→green, land by tag + cherry-pick | `In Progress` | `specs/in-progress/` |
13
13
  | `/spec-incident` | (Incident) Capture a live outage as a record written during it — grill, write in flight, fixes as separate linked specs | `In Progress` | `specs/in-progress/` |
@@ -157,6 +157,9 @@ build phase 1 now? Say yes and it carries on into a bare `/spec-next`; say no an
157
157
  you are already standing in the worktree, so `/spec-next` typed later does the
158
158
  same thing. The primary checkout stays on the base branch throughout — the
159
159
  session followed the spec, nothing was checked out anywhere.
160
+ That is also why the base branch reads `backlog` for a spec being built: the
161
+ move happened on its branch. **`skitterspec status`** reads each worktree's own
162
+ copy and lists what is in flight and how far along it is.
160
163
  **`checkout`** builds the branch in the primary checkout instead: one spec at a
161
164
  time, nowhere else to stand, and the build always carries straight on.
162
165
  The difference is how many specs can be in flight at once.
@@ -179,7 +182,7 @@ your canonical `localhost` ports so you can test it at the normal URL
179
182
  move, header edits, the code) happens on the spec's branch in the worktree; `main`
180
183
  changes only when it merges. Teardown is folded into `/spec-complete` ·
181
184
  `/spec-cancel`. Beneath the skills, `skitterspec spec-env
182
- <up|bootstrap|nospec|main|down|prune|dev|connect|integrate|hotfix|live|review|stage|status|resolve>` is the CLI
185
+ <up|bootstrap|nospec|main|down|prune|dev|connect|integrate|disposition|handoff|hotfix|live|review|stage|status|resolve|base>` is the CLI
183
186
  engine. `up` plans the fork;
184
187
  **`bootstrap` then runs the worktree's own steps itself** — seeding the gitignored files, the env file, `setup`, the stack — so
185
188
  the command a skill runs names no file, and a secrets hook that refuses any
@@ -234,6 +237,16 @@ worktree: `Commit & Start` keeps it and carries into `/spec-start`, plain
234
237
  `commit` tears it down, because a spec parked in the backlog should hold no
235
238
  checkout.
236
239
 
240
+ **`land.specDocs: "branch"` is the exception, taken deliberately.** Where the
241
+ base branch accepts no direct push even for documents, the spec rides its
242
+ feature branch and lands with the work's PR: both verdicts commit and skip the
243
+ integrate, plain `commit` keeps branch and worktree (they are the document's
244
+ only copy), and the backlog is found through the tracker listing rather than
245
+ `ls specs/backlog/` on base — the known branch-relative-bucket cost, accepted
246
+ because the alternative is a spec that cannot be written at all. The worktree
247
+ fork-point check is exempt in this mode for the hotfix's reason: the spec is
248
+ *supposed* to be absent from the fork point.
249
+
237
250
  **What this bought is the reason it changed.** The page waits for a verdict
238
251
  without a timeout — deliberately, because a reader walking away from a diff is
239
252
  the normal case — and while the spec was authored in the primary checkout that
@@ -497,15 +510,60 @@ against a committed last-pushed snapshot and applies it) and `/spec-status`
497
510
  backed by a `spec-sync` CLI. There is no content pull — the tracker is never
498
511
  read back or merged.
499
512
 
500
- **It tells, it doesn't track.** Skitterspec sends the tracker its own
501
- transitions — a spec starting, a phase finishing, a spec completing or being
502
- cancelled — and nothing else. A status someone changes on the tracker is
503
- theirs to have changed: it is never read back, never reported, and only
504
- overwritten when the spec itself next moves. Releases and deployments are the
505
- host project's to record. The one release question skitterspec answers is which
506
- tickets a commit range contains (`spec-sync released <range> --json`, each row
507
- with its spec and url), because only the `Refs:` trailers know. What the
508
- project's CI does with that list is its own business.
513
+ **What a ticket carries is the project's choice (`mapping.description`).**
514
+ Under `"spec"` the issue body is the whole overview and each phase sub-issue
515
+ its phase file, replaced on every push — what every config written before the
516
+ setting existed gets. Under `"summary"`, what a new config is written with, the
517
+ spec stays in git and the ticket gets what a ticket is for: a delimited
518
+ **block** holding the spec's `## Summary`, its `## Acceptance criteria`, the
519
+ Impact table's surfaces, its gating and its progress (`62% · phase 3 of 4`),
520
+ with each sub-issue's block holding its goal and task count. Push **splices**
521
+ the block into whatever the description holds, so a team's own text around it
522
+ survives every push, and only an edit *inside* the block is drift. That is a
523
+ splice, not a pull: nothing outside the block is parsed, merged or written
524
+ back. A delimiter edited by hand is cannot-tell — the push names it and writes
525
+ nothing there rather than guess. `mapping.phases: "none"` drops the phase
526
+ objects entirely; `subissue` stays the default because assignable phases are
527
+ what let them run in parallel.
528
+
529
+ **The tracker owns status; the repo owns the spec.** With a provider attached,
530
+ a team runs its work from the board — assignment, Ready, Blocked, Cancelled and
531
+ Done all happen there — so a status kept in the repo as well is a copy somebody
532
+ has to keep in step by hand, and on the base branch it was already wrong (an
533
+ in-flight spec still reads `backlog` there). The split follows what each side
534
+ can actually know:
535
+
536
+ 1. **The repo owns the content** — Problem, Decisions, phases, checkboxes,
537
+ Changelog. Content follows branches, lands in the same commit as its code,
538
+ and is read offline by the agent building it. That is why specs did not move
539
+ into the tracker wholesale: a ticket has no branches.
540
+ 2. **The tracker owns status and assignee** — `sync.fieldOwnership`
541
+ `workflowState` and `assignee` ship as **`pull`**. Under `pull` the skills
542
+ write no `Status:` header, no `Developer:` line and no State log row (the
543
+ `spec-status-owner` seam asks `spec-sync ownership` first), `assign` writes
544
+ the ticket directly, and the repo stores nothing it reads.
545
+ 3. **Buckets are code facts**, not planning status: `backlog` (not started),
546
+ `in-progress` (`/spec-start` ran), `complete` (landed — merged on the host
547
+ under PR landing), `cancelled` (torn down). Only skills that act on code move
548
+ them.
549
+ 4. **The repo writes status only when it acts on code** — creation, start,
550
+ complete, cancel — detected as a bucket move, never on a push that merely
551
+ cannot tell. Under PR landing the complete transition waits for run 2: a
552
+ ready PR is not a merged one.
553
+ 5. **A board that finished first is reported, never obeyed.** `spec-sync board`
554
+ says one line when the ticket is Done or Cancelled while the spec is still in
555
+ flight, naming the cleanup; `/spec-start` and `/spec-next` warn and carry on,
556
+ `/spec-status` reports. Nobody is obliged to sync, because nothing reads the
557
+ bucket as planning truth any more.
558
+
559
+ `"push"` on either field restores the repo-owned behaviour; `"none"` leaves it
560
+ alone entirely. Without a provider none of this applies: the header, the State
561
+ log and the buckets keep their meaning, because there is nowhere else for status
562
+ to live. Releases and deployments are the host project's to record. The one
563
+ release question skitterspec answers is which tickets a commit range contains
564
+ (`spec-sync released <range> --json`, each row with its spec and url), because
565
+ only the `Refs:` trailers know. What the project's CI does with that list is its
566
+ own business.
509
567
 
510
568
  **One-way means the repo wins, and it now wins *loudly*.** Those are different
511
569
  claims and only the first was ever in doubt. A ticket's description is generated
@@ -538,13 +596,14 @@ disk at all. The tracker knows both, and each row carries the spec's folder name
538
596
  so `/spec-start <name>` is a copy-paste away. See the provider's own docs for
539
597
  what it calls the command.
540
598
 
541
- A provider may also mirror **who is building a spec**, as one more field the repo
542
- owns. Where it does, `/spec-start` records the developer and the tracker's ticket
543
- is assigned to them, the assignment is released when the spec completes, and
544
- `/spec-claim` moves ownership mid-flight — take it, hand it back, or hand it to a
545
- teammate. Like everything else here it is opt-in and one-way: the spec file is
546
- the record, and the ticket is the mirror. See the provider's own docs for how to
547
- turn it on.
599
+ A provider also knows **who is building a spec**, and under the default `pull`
600
+ the ticket's assignee is the record: `/spec-start` assigns the operator only when
601
+ nobody holds the ticket (`assign --if-unassigned`), never reassigns anyone, and
602
+ leaves the assignee in place when the spec finishes — the board keeps who did
603
+ the work. `/spec-claim` moves ownership mid-flight — take it, hand it back, or
604
+ hand it to a teammate — by writing the ticket directly. Under `push` the spec
605
+ file is the record and the ticket the mirror, as before. See the provider's own
606
+ docs for both.
548
607
 
549
608
  **Every skill that moves a spec through the lifecycle carries a seam**, so the
550
609
  mirror keeps up without anyone remembering to push: `/spec`, `/spec-bug` and
@@ -713,6 +772,10 @@ Every spec header carries:
713
772
  (one-way; the repo owns the link), and `/spec-complete` on the incident
714
773
  counts the spec among its fixes. Absent on everything else.
715
774
 
775
+ **Under a provider's default `pull`, these three are the tracker's** —
776
+ `Status:`, `Developer:` and the State log — and new specs carry
777
+ `> **Tracker:** <url>` instead; see the ticketing section above. A spec that already has them keeps them as history.
778
+
716
779
  Every spec also has a **State log** table — the audit trail of folder/status
717
780
  transitions. Each lifecycle skill appends exactly one row when it changes state:
718
781
 
@@ -448,7 +448,7 @@ behind it — say nothing rather than reporting that there was nothing to report
448
448
  | `Built` | What the run produced — the code, the edits, the spec written. |
449
449
  | `Tests` | The result and the command that produced it. |
450
450
  | `Notes` | What this run hit and handled — one short paragraph. Named `Snags` once, which announced a problem before the reader had read one; most of what belongs here is neither good nor bad, just worth knowing. A wrong turn, a guard that fired, a check of yours that turned out to be a false negative. Not a caveat on the outcome (`⚠️` and `Why` carry those) and not future work (`Follow-ups` carries that): this is what happened on the way. |
451
- | `Landed` | A fast-forward, a tag, a cherry-pick. |
451
+ | `Landed` | A fast-forward, a tag, a cherry-pick — or in PR mode the PR itself: `PR #14 (draft) updated` · `PR #14 opened — finish after merge` · `PR #14 merged` · `held — not pushed` (a declined consent is a true outcome, not a failure). |
452
452
  | `Worktree` | A worktree provisioned, entered, or torn down. |
453
453
  | `Untouched` | Uncommitted work the run deliberately left alone — whose, and how much. |
454
454
  | `Review` | The rendered diff page: files, `+`/`−`, and the stack — the same tiers the banner lists, labelled, run together with `·` because a row is one cell, with any `network` alternates inline after the network URL (`also <url> (tailnet)`), plus `live` where the engine reported one. **No question** — a row cannot be waited on, so a question in one is unanswerable by construction (see *asking implies waiting*). **Omitted entirely when the run is waiting on a verdict**: the banner after the block carries it instead, and a row saying the same thing beside it splits the reader's attention across two places. |
@@ -157,7 +157,9 @@ not write the spec until this is resolved.
157
157
  asking. Verify endpoints/models/files actually exist before relying on them.
158
158
  - Cover, at minimum, the areas that apply:
159
159
  1. **Problem & why** — what's broken/missing, who feels it, why now.
160
- 2. **Scope & non-goals** — explicit out-of-scope items.
160
+ 2. **Scope & non-goals** — explicit out-of-scope items, and the
161
+ **acceptance criteria**: what someone checks, without reading the code,
162
+ to call the work done (they become `## Acceptance criteria`).
161
163
  3. **Affected areas** — concrete files/modules/packages this touches.
162
164
  4. **Proposed solution shape** — the chosen approach and the alternatives
163
165
  rejected, with the reason (this becomes "Decisions").
@@ -330,6 +332,28 @@ move to, **warn, don't refuse**, and say the trailer for that commit wants
330
332
  a feature, `chore-` for a chore, `issue-` for an issue (the bug counterpart
331
333
  uses `bug-`, the incident one `incident-`).
332
334
 
335
+ **Who owns this spec's status.** Only when `specs/.core/linear.config.json` exists.
336
+ Run `skitterspec spec-sync ownership --json` once and read both values:
337
+
338
+ - **`workflowState: "pull"`** — Linear is the record of this spec's status, so
339
+ this skill **writes no `> **Status:**` header and appends no State log row**,
340
+ wherever its steps below say to. A spec this skill creates gets neither: its
341
+ header carries `> **Tracker:** <issue url>` instead, filled in once linking
342
+ has stamped the url, and its body has no `## State log` section. Readiness is
343
+ the spec's own content — a `Draft` is a spec whose **Open questions** is not
344
+ empty — so nothing is lost with the header.
345
+ - **`assignee: "pull"`** — Linear holds who is building it, so this skill
346
+ **writes no `> **Developer:**` line** either.
347
+ - **Anything else** — `push`, `none`, sync reported off, or the command failing
348
+ — is cannot-tell, and the steps below run exactly as written.
349
+
350
+ What `pull` leaves alone: the folder move (a bucket records what the repo did to
351
+ the code, not a planning status), the phase files and index, and the Changelog.
352
+ A reason a `Status:` header would have carried — a cancellation's, a knock-back
353
+ to Draft — goes in a dated Changelog line instead. A spec that already carries
354
+ any of these lines keeps them untouched: they are history, and nothing rewrites
355
+ them.
356
+
333
357
  Use this template (keep it **as concise as possible** — no filler, no restating
334
358
  the codebase, link rather than duplicate):
335
359
 
@@ -354,10 +378,20 @@ the codebase, link rather than duplicate):
354
378
  > "none: not user-facing". An empty value or a bare
355
379
  > "none" is not a valid outcome — see Phase A item 10>
356
380
 
381
+ ## Summary
382
+
383
+ <1–3 sentences: what the work delivers, for someone who will never open the
384
+ spec — no phase breakdown, no implementation detail. With a ticketing provider
385
+ in summary mode this is what the ticket leads with.>
386
+
357
387
  ## Problem
358
388
 
359
389
  <2–6 sentences: what's wrong/missing and why it matters. No fluff.>
360
390
 
391
+ ## Acceptance criteria
392
+
393
+ - <an observable check, from Phase A item 2 — one per line>
394
+
361
395
  ## Decisions
362
396
 
363
397
  <Numbered, confirmed decisions from Phase A. Each: the choice + one-line why,
@@ -467,6 +501,11 @@ Rules for the spec body:
467
501
  - **Changelog** is mandatory and lives in the spec. Every later decision or
468
502
  course-correction gets a dated one-line entry. Convert relative dates to
469
503
  absolute.
504
+ - **`## Summary` and `## Acceptance criteria` are the ticket-facing pair.**
505
+ Write them for a reader on the board, not the one building it: the summary
506
+ says what the work delivers, each criterion is something a person can check.
507
+ A ticketing provider in summary mode projects exactly these (plus the
508
+ Impact table's surfaces, Gating and progress) and nothing else.
470
509
  - Keep it tight. If a section adds no information, delete it.
471
510
 
472
511
  ## Phase C — finish up
@@ -550,6 +589,18 @@ on the branch you are standing on — so a spec left on its own branch would be
550
589
  invisible to everyone, including the next `/spec` run. The commit goes on the
551
590
  branch and then the branch goes onto the base:
552
591
 
592
+ **Unless the project set `land.specDocs: "branch"`** — the lane for a base
593
+ branch nobody may push, where the spec document rides the feature branch and
594
+ lands with the work's PR. There both verdicts commit on the branch and
595
+ **skip the integrate below entirely**; `commit-start` carries into
596
+ `/spec-start` over the same worktree exactly as ever, and plain `commit` keeps
597
+ the branch AND the worktree, because they are now the only copy of the
598
+ document. Say the trade out loud in the report's `Notes`: the backlog is found
599
+ through the tracker listing (or `git branch`), not `ls specs/backlog/` on base
600
+ — and a later `/spec-cancel` of a never-started branch-mode spec deletes the
601
+ branch and worktree with the spec document dying with them, no PR ever having
602
+ existed. Under the default `"trunk"` everything below applies unchanged.
603
+
553
604
  ```
554
605
  skitterspec spec-env integrate <name>
555
606
  ```
@@ -671,7 +722,9 @@ If it's absent, skip this entirely — the spec stays local-only and the skill
671
722
  behaves exactly as above. When present, link the spec you just wrote so status
672
723
  and discussion live in Linear while the repo stays the source of truth. A spec is
673
724
  a Linear **issue**; each phase is a **sub-issue**, carrying that phase's tasks in
674
- its description as a read-only checklist.
725
+ its description as a read-only checklist. That is the `spec` description mode;
726
+ under `summary` (a new config's default) the issue and each sub-issue carry
727
+ a short block instead, and the phase's tasks stay in git.
675
728
 
676
729
  **Linking is just the first push**, so it runs the same engine path `/spec-push`
677
730
  does — you never create the issue by hand:
@@ -291,6 +291,28 @@ split it into phase files (`01-<slug>.md`, `02-…`) with a phase index in
291
291
  `00-overview.md`, exactly like a feature spec. It starts in `in-progress`
292
292
  because work is already underway. Keep it lean:
293
293
 
294
+ **Who owns this spec's status.** Only when `specs/.core/linear.config.json` exists.
295
+ Run `skitterspec spec-sync ownership --json` once and read both values:
296
+
297
+ - **`workflowState: "pull"`** — Linear is the record of this spec's status, so
298
+ this skill **writes no `> **Status:**` header and appends no State log row**,
299
+ wherever its steps below say to. A spec this skill creates gets neither: its
300
+ header carries `> **Tracker:** <issue url>` instead, filled in once linking
301
+ has stamped the url, and its body has no `## State log` section. Readiness is
302
+ the spec's own content — a `Draft` is a spec whose **Open questions** is not
303
+ empty — so nothing is lost with the header.
304
+ - **`assignee: "pull"`** — Linear holds who is building it, so this skill
305
+ **writes no `> **Developer:**` line** either.
306
+ - **Anything else** — `push`, `none`, sync reported off, or the command failing
307
+ — is cannot-tell, and the steps below run exactly as written.
308
+
309
+ What `pull` leaves alone: the folder move (a bucket records what the repo did to
310
+ the code, not a planning status), the phase files and index, and the Changelog.
311
+ A reason a `Status:` header would have carried — a cancellation's, a knock-back
312
+ to Draft — goes in a dated Changelog line instead. A spec that already carries
313
+ any of these lines keeps them untouched: they are history, and nothing rewrites
314
+ them.
315
+
294
316
  ```markdown
295
317
  # Bug: <short title>
296
318
 
@@ -369,7 +391,9 @@ If it's absent, skip this entirely — the spec stays local-only and the skill
369
391
  behaves exactly as above. When present, link the spec you just wrote so status
370
392
  and discussion live in Linear while the repo stays the source of truth. A spec is
371
393
  a Linear **issue**; each phase is a **sub-issue**, carrying that phase's tasks in
372
- its description as a read-only checklist.
394
+ its description as a read-only checklist. That is the `spec` description mode;
395
+ under `summary` (a new config's default) the issue and each sub-issue carry
396
+ a short block instead, and the phase's tasks stay in git.
373
397
 
374
398
  **Linking is just the first push**, so it runs the same engine path `/spec-push`
375
399
  does — you never create the issue by hand:
@@ -450,6 +474,12 @@ is a cosmetic gap that `/spec-claim` closes later.
450
474
  skitterspec spec-sync assign <spec> --to <user-id> --name "<display name>"
451
475
  ```
452
476
 
477
+ **Where `spec-sync ownership` reports `assignee: pull`**, the issue is the
478
+ record and the spec file is not: add `--if-unassigned`. It writes Linear
479
+ directly, claims the issue only when nobody holds it, and stamps nothing —
480
+ so step 4 has no header to leave alone, since under `pull` no `Developer:`
481
+ line was written.
482
+
453
483
  4. **Leave `> **Developer:**` as `git config user.name`** — the step above this
454
484
  seam already set it, and this seam must not overwrite it with the tracker's
455
485
  display name.
@@ -471,12 +501,15 @@ is a cosmetic gap that `/spec-claim` closes later.
471
501
 
472
502
  **Nothing is pushed here.** `assign` writes the repo only, and the refresh these
473
503
  skills already run sends it. Assignment is an ordinary field of the projection,
474
- not a side errand with its own network call.
504
+ not a side errand with its own network call. Under `pull` the
505
+ `assign` call is itself the write, so there is nothing for a push to carry.
475
506
 
476
507
  **There is no unassign step anywhere.** The projection derives the assignee from
477
508
  the spec's lifecycle bucket, so `/spec-complete` and `/spec-cancel` release the
478
509
  issue through the push they already make. The stamp deliberately stays in the
479
- file: who actioned the work outlives who is currently holding it.
510
+ file: who actioned the work outlives who is currently holding it. Under `pull`
511
+ the projection carries no assignee at all, so finishing a spec leaves it on
512
+ the issue: the board keeps who did the work.
480
513
 
481
514
  ### Picking the Linear Project
482
515
 
@@ -37,6 +37,28 @@ it verbatim/condensed for the header.
37
37
 
38
38
  ## 4. Stamp the spec
39
39
 
40
+ **Who owns this spec's status.** Only when `specs/.core/linear.config.json` exists.
41
+ Run `skitterspec spec-sync ownership --json` once and read both values:
42
+
43
+ - **`workflowState: "pull"`** — Linear is the record of this spec's status, so
44
+ this skill **writes no `> **Status:**` header and appends no State log row**,
45
+ wherever its steps below say to. A spec this skill creates gets neither: its
46
+ header carries `> **Tracker:** <issue url>` instead, filled in once linking
47
+ has stamped the url, and its body has no `## State log` section. Readiness is
48
+ the spec's own content — a `Draft` is a spec whose **Open questions** is not
49
+ empty — so nothing is lost with the header.
50
+ - **`assignee: "pull"`** — Linear holds who is building it, so this skill
51
+ **writes no `> **Developer:**` line** either.
52
+ - **Anything else** — `push`, `none`, sync reported off, or the command failing
53
+ — is cannot-tell, and the steps below run exactly as written.
54
+
55
+ What `pull` leaves alone: the folder move (a bucket records what the repo did to
56
+ the code, not a planning status), the phase files and index, and the Changelog.
57
+ A reason a `Status:` header would have carried — a cancellation's, a knock-back
58
+ to Draft — goes in a dated Changelog line instead. A spec that already carries
59
+ any of these lines keeps them untouched: they are history, and nothing rewrites
60
+ them.
61
+
40
62
  Update the **Status** header in the entry point so the reason is visible at the
41
63
  top:
42
64
 
@@ -74,6 +96,10 @@ Linear API key is set (see `apply.transport` in `linear.config.md`).
74
96
  and **finish the operation anyway**. The spec is complete/cancelled/reviewed in
75
97
  the repo regardless; the mirror is disposable and the next push repairs it.
76
98
  Do not roll anything back, and do not stop to ask.
99
+ - **In `/spec-complete`'s PR mode this push belongs to run 2.** Run 1 readies
100
+ a PR, and a ready PR is not a merged one — so skip it on run 1, even where
101
+ the `branch` lane committed the move to ride the PR, and run it on run 2 once
102
+ the host has merged. The tracker's Done lands when the work does.
77
103
  - **Say what happened** in the skill's report: mirror updated, skipped as
78
104
  unlinked, or failed with the reason.
79
105
  - **There is no unassign step, and that is not an omission.** Where the repo owns
@@ -82,6 +108,8 @@ Linear API key is set (see `apply.transport` in `linear.config.md`).
82
108
  releases the issue, and the push above carries it like any other field. The
83
109
  `linear_assignee_id` stamp and the `> **Developer:**` header both stay put:
84
110
  they record who actioned the work, which outlives who is holding it.
111
+ Under `assignee: pull` the projection carries no assignee at all, so the
112
+ issue keeps it: the board records who did the work.
85
113
 
86
114
  ### Why it sits here
87
115
 
@@ -254,6 +282,15 @@ only ordering that survives.
254
282
  copy of abandoned work. Seeing no such section here is the expected case, not
255
283
  a fault — and if the user wants the remote branch gone anyway, that is their
256
284
  call to make explicitly, not something to tidy away on their behalf.
285
+
286
+ **In PR mode the same confirm-first section may carry the PR abandon**
287
+ (`gh pr close` / `az repos pr update --status abandoned`): a cancelled spec
288
+ must not leave a live PR behind — it is noise with a merge button on it.
289
+ The teardown of an open-PR spec is refused unforced (the PR is named), so
290
+ this line appears on the forced run the refusal led to; ask before running
291
+ it like the delete above, and run the abandon **before** any remote branch
292
+ delete — closing the PR first is what stops the host complaining about a
293
+ deleted source branch.
257
294
  4. `skitterspec spec-env prune` — reap orphaned test-DB volumes that belong to no
258
295
  live spec (leftovers from declined/aborted teardowns or manual worktree
259
296
  removal). Show the orphan list and, **only on the user's confirmation**, run
@@ -28,6 +28,27 @@ thing enforcing it for long enough that the engine's own success line —
28
28
  `next: push it, so Linear agrees` — was printed over a push that would drop the
29
29
  field. Prose in one skill is not a guard; it is a guard's documentation.
30
30
 
31
+ ## Under `pull`, the issue is the record
32
+
33
+ Ask once: `skitterspec spec-sync ownership --json`. Where `assignee` is
34
+ **`pull`**, Linear owns who is building the spec, and the repo keeps no copy — so
35
+ the steps below change in four places, and only these:
36
+
37
+ - **`assign` writes Linear directly and stamps nothing.** There is no repo record
38
+ to change and no push to follow: §4 does not run, because the `assign` call
39
+ *was* the write.
40
+ - **Take checks Linear, not the file.** Run `assign <spec> --to <id> --if-unassigned`
41
+ first. If it reports the issue is already assigned to someone else, that is
42
+ step 2's colleague: name them, confirm, then run it again without the flag.
43
+ - **No `> **Developer:**` line is written**, by any mode. The tracker names the
44
+ person; a header would be the copy `pull` exists to remove.
45
+ - **§3 is skipped.** The issue's own history records the move, and a
46
+ Changelog line saying who holds it would be that copy again.
47
+
48
+ Everything else stands: the opt-in check, resolving a person through a search
49
+ before a hand-over, and the report. Under `push` (or with ownership unknown)
50
+ nothing in this section applies.
51
+
31
52
  ## 1. Identify the target spec
32
53
 
33
54
  The argument, else the spec in flight for this session, else ask. Unlike
@@ -78,8 +99,26 @@ now".
78
99
  stranger and nothing downstream would notice.
79
100
  2. On several matches, show them and let the user pick. On none, say so and
80
101
  stop — do not fall back to the raw argument as an id.
81
- 3. Stamp it with the resolved id and display name.
82
- 4. Set `> **Developer:**` to that person's **Linear display name**.
102
+ 3. **Make the work reachable before you hand it over.** Run
103
+ `skitterspec spec-env handoff <spec> --json`. A spec's branch lives in a
104
+ local worktree, so without this the issue would name a new owner for code
105
+ only you can see.
106
+ - **`blocked`** — relay the reason. Where it carries an `offer` (uncommitted
107
+ work: `/commit`), offer it once as `offered-unblocks.md` says, run it on a
108
+ yes, then run `handoff` again. With no offer, stop here: nothing is assigned.
109
+ - **`commands`** non-empty — the branch is not published at HEAD. Ask once,
110
+ naming the target: *"push `<branch>` to `<remote>` so <name> can pick it
111
+ up?"* Run it on a yes. On a no, stop: a hand-over the new owner cannot
112
+ reach is not a hand-over, so nothing is assigned.
113
+ - Then post the note it wrote onto the issue:
114
+ `skitterspec spec-sync comment <spec> --file <notePath>`. It names the
115
+ branch, the sha and the phases done, and a comment is the surface no later
116
+ push replaces. A failed post is said and does not stop the hand-over.
117
+
118
+ The new owner's `/spec-start <spec>` resumes from that branch rather than
119
+ forking a fresh one (`spec-env up` sees it on the remote).
120
+ 4. Stamp it with the resolved id and display name.
121
+ 5. Set `> **Developer:**` to that person's **Linear display name**.
83
122
 
84
123
  This is the **one** place the tracker's name reaches the header, and the
85
124
  exception is the whole reason it exists: there is no `git config user.name`
@@ -102,6 +102,29 @@ feature was designed not to disturb.
102
102
 
103
103
  - Tick all completed tasks in the phase files; flip every finished phase-file
104
104
  heading **and** every row in the `00-overview.md` phase index to `✅`.
105
+
106
+ **Who owns this spec's status.** Only when `specs/.core/linear.config.json` exists.
107
+ Run `skitterspec spec-sync ownership --json` once and read both values:
108
+
109
+ - **`workflowState: "pull"`** — Linear is the record of this spec's status, so
110
+ this skill **writes no `> **Status:**` header and appends no State log row**,
111
+ wherever its steps below say to. A spec this skill creates gets neither: its
112
+ header carries `> **Tracker:** <issue url>` instead, filled in once linking
113
+ has stamped the url, and its body has no `## State log` section. Readiness is
114
+ the spec's own content — a `Draft` is a spec whose **Open questions** is not
115
+ empty — so nothing is lost with the header.
116
+ - **`assignee: "pull"`** — Linear holds who is building it, so this skill
117
+ **writes no `> **Developer:**` line** either.
118
+ - **Anything else** — `push`, `none`, sync reported off, or the command failing
119
+ — is cannot-tell, and the steps below run exactly as written.
120
+
121
+ What `pull` leaves alone: the folder move (a bucket records what the repo did to
122
+ the code, not a planning status), the phase files and index, and the Changelog.
123
+ A reason a `Status:` header would have carried — a cancellation's, a knock-back
124
+ to Draft — goes in a dated Changelog line instead. A spec that already carries
125
+ any of these lines keeps them untouched: they are history, and nothing rewrites
126
+ them.
127
+
105
128
  - Set the **Status** header in the entry point:
106
129
  `> **Status:** Complete (<YYYY-MM-DD>)`.
107
130
  - Append a **State log** row:
@@ -135,6 +158,10 @@ Linear API key is set (see `apply.transport` in `linear.config.md`).
135
158
  and **finish the operation anyway**. The spec is complete/cancelled/reviewed in
136
159
  the repo regardless; the mirror is disposable and the next push repairs it.
137
160
  Do not roll anything back, and do not stop to ask.
161
+ - **In `/spec-complete`'s PR mode this push belongs to run 2.** Run 1 readies
162
+ a PR, and a ready PR is not a merged one — so skip it on run 1, even where
163
+ the `branch` lane committed the move to ride the PR, and run it on run 2 once
164
+ the host has merged. The tracker's Done lands when the work does.
138
165
  - **Say what happened** in the skill's report: mirror updated, skipped as
139
166
  unlinked, or failed with the reason.
140
167
  - **There is no unassign step, and that is not an omission.** Where the repo owns
@@ -143,6 +170,8 @@ Linear API key is set (see `apply.transport` in `linear.config.md`).
143
170
  releases the issue, and the push above carries it like any other field. The
144
171
  `linear_assignee_id` stamp and the `> **Developer:**` header both stay put:
145
172
  they record who actioned the work, which outlives who is holding it.
173
+ Under `assignee: pull` the projection carries no assignee at all, so the
174
+ issue keeps it: the board records who did the work.
146
175
 
147
176
  ### Why it sits here
148
177
 
@@ -290,6 +319,27 @@ in that order, and stops at the first step that fails:
290
319
  **Then stop: the worktree is kept** (step 7). The branch is what the next
291
320
  `-hotfix.N` is cut from if this deploy fails, so it waits for the deploy.
292
321
 
322
+ **A branch-based hotfix** (`> **Base version:** <branch>@<sha>`, from
323
+ `hotfix.releaseBranch`) runs the same command and the same consent ask — the
324
+ engine picks the path from the header. What differs is what it does:
325
+
326
+ - it **rebases** onto `<remote>/<branch>` and **lands on the release branch** by
327
+ `land.strategy`: under `ff` the push of the hotfix onto that branch *is* the
328
+ land (so the consent ask names the release branch, not only a tag); under `pr`
329
+ it pushes the hotfix branch and opens a **ready** PR, then holds with
330
+ `the PR into <branch> is open` — re-run the same command once it merges;
331
+ - the **tag** is optional (`hotfix.tag`): cut and pushed after the land, or not
332
+ at all;
333
+ - the **carry** goes to the resolved base — the current rc under
334
+ `baseBranchPattern` — directly under `ff`, through a `carry/<slug>` branch and
335
+ its own ready PR under `pr`.
336
+
337
+ `Landed` then names the release branch, how it landed (fast-forward, or the PR
338
+ and whether it merged), the tag if one was cut, and the carry. A PR still
339
+ waiting on its merge is a `⚠️` — the land is not done until the host says so.
340
+ `spec-env hotfix finish` tears it down once both the land and the carry are
341
+ confirmed, naming every check that is not.
342
+
293
343
  ### 6-feature/bug. `Type: Feature` / `Type: Bug` — rebase + fast-forward
294
344
 
295
345
  Land the finished branch on the base branch so the work reaches `main` (or your
@@ -325,6 +375,19 @@ is finished — e.g. to run a later phase in CI or a shared test env? Use
325
375
  `git -C <worktree> rebase --abort`, relay the conflict, and **stop** — leave it
326
376
  to the user; do not tear anything down.
327
377
  On a **no-op** ("already landed"), just say so and continue.
378
+
379
+ **Under a rolling base** (`baseBranchPattern` in `env.config.json`) the plan
380
+ can carry two more things, and both are the engine's to say:
381
+ - a `git -C <mainRepoPath> switch <base>` (or `switch -c … --track`) **first**,
382
+ when the primary checkout still has an older release candidate out — run it
383
+ in order with the rest;
384
+ - a `retarget: forked from <old> · landing onto <new>` line, when the rc
385
+ rolled while this spec was in flight. It is the land working, not a fault:
386
+ relay it verbatim in the `Landed` row so the reader sees where the work went.
387
+
388
+ A refusal naming `no branch matches <pattern>` means no release candidate is
389
+ visible — usually not fetched. Relay it, stop with `⏸`, and name
390
+ `skitterspec spec-env base` (it fetches and lists) as the next step.
328
391
  3. **Run the after-land commands.** Where the plan ends with a
329
392
  `then, in the primary checkout (<path>), run:` section — the project's
330
393
  `afterLand`, such as a data migration — run those commands in that
@@ -344,6 +407,58 @@ is finished — e.g. to run a later phase in CI or a shared test env? Use
344
407
  5. **Report** the landing (base branch, fast-forward result). It **never pushes** —
345
408
  mention the user can `git push` the base branch themselves.
346
409
 
410
+ ### 6-pr. PR mode (`land.strategy: "pr"`) — dispose, push, ready; finish after the merge
411
+
412
+ Completion is **two runs** here, the `hotfix land`/`finish` shape: the merge is
413
+ the host's act and may be days away, so the run that readies the PR cannot also
414
+ finalise. Which run this is, the engine says: `spec-env integrate <name>`
415
+ printing a `(pr)` plan is run 1; printing
416
+ `already landed … — finish what the merge deferred` is run 2.
417
+
418
+ **Run 1 — the last control point.** Once the PR is ready, anyone can merge from
419
+ the host UI and the tracker's automations act — so everything that needs a
420
+ person happens before the ready, in this order:
421
+
422
+ 1. **Disposition pass — mandatory, before anything is pushed.** Run
423
+ `skitterspec spec-env disposition <name> --json`. For every unticked task,
424
+ non-✅ phase and surfaced follow-up it lists, the user disposes it:
425
+ **tick** (it was done — tick it, naming the evidence) · **spin off** (a
426
+ follow-on ticket minted now, linked to this one — offer `/spec` or the
427
+ provider's mint) · **drop** (a dated Changelog line with the reason, the
428
+ `none: <reason>` grammar). Batch the asking; never decide for them, and
429
+ never auto-create. **Refuse to ready the PR while anything is undisposed**
430
+ — you asked the enumeration, so it cannot also be fine. `clean` proceeds
431
+ with nothing asked.
432
+ 2. **Consent, naming the target.** Ask once, plainly:
433
+ *"push `<branch>` and open the PR into `<base>` on `<remote>`, ready for
434
+ merge?"* — the base branch is named so a spec that outlived a release roll
435
+ is caught by the person reading a stale rc name. On a no, stop with the
436
+ held commands relayed.
437
+ 3. **Run the plan.** `skitterspec spec-env integrate <name> --push` and run
438
+ what it prints: fetch, rebase onto `<remote>/<base>`, the push, and the PR
439
+ create — or, where a draft already exists, the flip-to-ready. A rebase
440
+ conflict aborts and stops exactly as 6-feature/bug step 2.
441
+ 4. **Defer the finalise.** §4's bucket move follows the spec documents'
442
+ `land.specDocs` lane: under `trunk` leave the spec in `specs/in-progress/`
443
+ — run 2 moves it on base after the merge; under `branch` commit the move on
444
+ the branch *before* step 3 so it rides the PR. Either way
445
+ **the tracker's Done waits for run 2**: a ready PR is not a merged one.
446
+ 5. **Stop, keeping the worktree.** Report
447
+ `Landed: PR #N opened — finish after merge` (or `ready`), spec
448
+ `In Progress`, worktree standing. No afterLand runs — the plan deferred it
449
+ — and no teardown. The merge is what un-blocks run 2; nothing here waits on
450
+ it.
451
+
452
+ **Run 2 — after the host merges.** `integrate` recognises the landed branch
453
+ (ancestry or the host's answer) and prints the finish: catch local base up and
454
+ the deferred `afterLand` with the land's own range. Run those, then do what run
455
+ 1 deferred — the §4 move (trunk lane) with its State log row and commit, the
456
+ §5 report — then **§7 teardown runs as normal**:
457
+ a host-merged branch counts as landed (squash and rebase
458
+ merges included), and a host that auto-deleted the branch makes the remote
459
+ delete an already-gone no-op, never a failure. Re-test on base before calling
460
+ it done, exactly as 6-feature/bug step 4.
461
+
347
462
  ## 7. Tear down the environment (opt-in, only if configured)
348
463
 
349
464
  **A hotfix is torn down later, by `hotfix finish`.**