feat: Chip-on-Board (CoB) packaging support (#259) - #265
Merged
Conversation
Design for issue #259: editable boolean Project.chip_on_board, a snapshot field on ManufacturabilityCheck, --cob precheck wiring (the image already supports it), and a service-layer re-check on toggle (TriggerReason.COB_CHANGE). Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Spec reviewer flagged a false 'slot/DRC-version snapshot' precedent: do_starting reads slot_size/full_id live from check.project, not a snapshot. Drop the proposed ManufacturabilityCheck.chip_on_board snapshot field and read --cob live from check.project (consistent with --slot/--id); fix service class reference. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Per review feedback, CoB toggling now mirrors the DRC_UPDATE/precheck-version mechanism instead of the file-replacement cancel path: a new model method create_check_cob_change() (parallel to create_check_drc_update) creates a new PENDING check with parent_check linkage and TriggerReason.COB_CHANGE; the existing _cancel_superseded_checks() auto-cancels any in-progress older check. Pure ORM, so no service/task-import layer is needed. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Spec review found _cancel_superseded_checks() is only called from checks_cleanup(), which is NOT in CELERY_BEAT_SCHEDULE, so the relied-upon auto-cancel never runs. DRC updates never hit this because they only re-check FINISHED checks. Since a CoB toggle can happen mid-check, create_check_cob_change() now marks an in-progress latest check as CANCELLING explicitly (mark_cancelling), relying only on the scheduled checks_pending/checks_cancelling pollers. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Spec reviewer (approved) flagged that the existing create_check_drc_update docstring repeats the unscheduled-cleanup belief; capture as a follow-up so it is not copied into create_check_cob_change. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Assuming PR #262 lands, checks_cleanup_superseded runs on a 60s beat and _cancel_superseded_checks/checks_cleanup no longer exist. Update the spec's rationale: the explicit cancel in create_check_cob_change() is kept for immediacy (a superseded check must not finish with a stale CoB flag), audit-precise cancel reason, and test determinism; the scheduled cleanup is demoted to a defence-in-depth backstop. Drop the now-obsolete docstring follow-up (fixed by #262) and note the rebase-after-#262 dependency. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Spec review findings: checks_cancelling only transitions CANCELLING->CANCELLED; container removal is done by checks_cleanup_orphaned_docker (60s) once the check is CANCELLED — correct all four passages that attributed teardown to checks_cancelling. Recast the explicit-cancel justification in terms of legal state transitions (ANALYZING->FINISHED vs CANCELLING->CANCELLED only) instead of the imprecise 'active' invariant, and add the detail-page CoB badge to the test plan. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
8 TDD tasks: model field + migration, COB_CHANGE trigger + migration, create_check_cob_change(), --cob command wiring, form + template, view re-check hook, detail badge, final verification. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Make Task 1's editability test fetch fresh from the DB so it genuinely fails before the field exists; give Task 5's form tests their own user/project (TestProjectForm.setUp only creates a shuttle) and correct both tasks' expected-failure descriptions; fix the create_check_drc_update end-line reference. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Add boolean USER_FIELD for Chip-on-Board packaging opt-in. Defaults to False, always user-editable (not subject to core-field immutability). Migration adds the field to both project and historicalproject tables. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Extends ManufacturabilityCheck.TriggerReason with COB_CHANGE ("cob_change")
to support re-triggering checks when a project's chip_on_board option changes.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Task 2's quality review found that _file_display.html's trigger-reason badge chains have no else fallback (a COB_CHANGE check would render no badge) and manufacturability_check_status.html falls back to a generic label. Add cob_change branches to all four chains plus a check-history badge test. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Add ManufacturabilityCheck.create_check_cob_change() which creates a PENDING check with TriggerReason.COB_CHANGE, explicitly cancels any in-progress source check (unlike DRC update which relies on superseded- check cleanup), and guards against being called on a non-latest check. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Append --cob to the Docker precheck command in do_starting when project.chip_on_board is True; add TDD test confirming the flag is present (cob=True) and absent (cob=False via existing regression test). Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Add chip_on_board field to ProjectForm.Meta.fields/widgets/help_texts and render it in project_form.html after is_public. The field is a user field (not a CORE_FIELD) so it stays enabled for non-staff on existing projects. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Quality review polish: drop the Meta.help_texts duplicate (ModelForm inherits the model field's help_text) and trim an overstated test docstring. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Add form_valid override on ProjectUpdateView that detects chip_on_board changes via form.changed_data and calls create_check_cob_change() on the latest check, cancelling any in-progress check and queuing a new PENDING COB_CHANGE check for the scheduled dispatcher to pick up. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Quality review fixes: replace the len([check, 'new']) lint-evasion idiom with EXPECTED_CHECKS_AFTER_COB_TOGGLE in tests/constants.py, annotate the _make_submitted_check helper, and derive the form payload's name/description from setUp's project instead of duplicating the strings. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Add Chip-on-Board badge to project detail page and CoB Change trigger-reason badges to all four if/elif chains in _file_display.html and manufacturability_check_status.html. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Quality review polish: assert the unambiguous 'Chip-on-Board (CoB)' string in the positive test, and make the standard-packaging test assert the Packaging block and Standard badge actually render rather than only absence. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: Organization UI Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (5)
🚧 Files skipped from review as they are similar to previous changes (4)
WalkthroughThis PR implements Chip-on-Board (CoB) packaging support by adding an editable ChangesChip-on-Board Packaging Feature
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~25 minutes Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 4
🧹 Nitpick comments (1)
wafer_space/projects/tests/test_forms.py (1)
92-101: ⚡ Quick winUse factories instead of direct ORM object creation in tests.
Line [92] and Line [95] create test data with
User.objects.create_user(...)and
Project.objects.create(...). Please use factory-boy here to keep test setup
consistent with the suite’s test-data contract.♻️ Proposed refactor
+from wafer_space.projects.tests.factories import ProjectFactory +from wafer_space.users.tests.factories import UserFactory @@ - user = User.objects.create_user( - username="formuser", email="form@example.com", password=TEST_PASSWORD - ) - project = Project.objects.create( - user=user, - name="Form Project", - shuttle=self.shuttle, - project_id="FRMP", - ) + user = UserFactory(username="formuser", email="form@example.com") + project = ProjectFactory( + user=user, + name="Form Project", + shuttle=self.shuttle, + project_id="FRMP", + )As per coding guidelines, "Use factory-boy for creating test data, not pytest fixtures."
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@wafer_space/projects/tests/test_forms.py` around lines 92 - 101, Replace direct ORM creation in the test with factory usage: instead of User.objects.create_user(...) and Project.objects.create(...), use the project's factory classes (e.g., UserFactory and ProjectFactory) to create the user and project, passing the same attributes (username, email, password/TEST_PASSWORD via factory params, and shuttle and project_id to ProjectFactory) and then instantiate ProjectForm with the factory-created user and instance; update references to User, Project, ProjectForm, TEST_PASSWORD, UserFactory and ProjectFactory accordingly so test data follows the suite’s factory conventions.Source: Coding guidelines
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@wafer_space/projects/models.py`:
- Around line 2330-2343: The create_check_cob_change method performs unlocked
reads and writes causing race conditions; fix it by wrapping the critical
section in a transaction.atomic block and reloading the key row with a
select_for_update lock (e.g., re-fetch self.project_file or the related
ManufacturabilityCheck via
project_file.latest_manufacturability_check.select_for_update()) before
re-evaluating latest_manufacturability_check and is_cancellable, then call
mark_cancelling and ManufacturabilityCheck.objects.create only after the locked
re-check to ensure a single authoritative decision and prevent duplicate
COB_CHANGE checks.
In `@wafer_space/projects/tests/test_views.py`:
- Around line 208-225: The test
test_detail_shows_cob_change_badge_in_check_history currently only creates one
ManufacturabilityCheck (via ManufacturabilityCheckFactory) so the COB_CHANGE
entry may be treated as the first entry and not rendered by the history branch
in _file_display.html; update the test to create an earlier (older) check for
the same project_file (e.g., a finished check with a different trigger_reason or
earlier timestamp) before creating the COB_CHANGE check so that the COB_CHANGE
check is a non-first history entry and the history-path rendering is exercised.
- Around line 499-512: The test test_unchanged_cob_creates_no_check currently
ignores the POST response so it can false-pass on form validation errors; after
calling self.client.post(url, self._cob_form_data(chip_on_board=False)) capture
the response, assert its status_code indicates success (e.g., 302 for redirect
after successful POST or 200 if no redirect), then refresh the related object
(use the project_file returned by self._make_submitted_check or reload it from
the DB) and assert its chip_on_board/project flag remains unchanged, and finally
assert
ManufacturabilityCheck.objects.filter(project_file=check.project_file).count()
== 1 to ensure no new check was created.
In `@wafer_space/projects/views.py`:
- Around line 257-261: After saving the project, calling
latest_check.create_check_cob_change() can raise ValueError and cause a 500;
wrap that call in a try/except that catches ValueError (only) and swallow or log
the error so the successful save isn't rolled back or bubbled as a 500. Locate
the call to latest_manufacturability_check.create_check_cob_change() in the view
(where cob_changed is handled), wrap it in try/except ValueError as e, and use
the view/logger to record a warning with the exception message while allowing
form_valid to return normally.
---
Nitpick comments:
In `@wafer_space/projects/tests/test_forms.py`:
- Around line 92-101: Replace direct ORM creation in the test with factory
usage: instead of User.objects.create_user(...) and Project.objects.create(...),
use the project's factory classes (e.g., UserFactory and ProjectFactory) to
create the user and project, passing the same attributes (username, email,
password/TEST_PASSWORD via factory params, and shuttle and project_id to
ProjectFactory) and then instantiate ProjectForm with the factory-created user
and instance; update references to User, Project, ProjectForm, TEST_PASSWORD,
UserFactory and ProjectFactory accordingly so test data follows the suite’s
factory conventions.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
Run ID: 3e071124-a98f-4a6e-9533-1d2233cd0133
📒 Files selected for processing (17)
docs/superpowers/plans/2026-06-10-chip-on-board-packaging.mddocs/superpowers/specs/2026-06-08-chip-on-board-packaging-design.mdwafer_space/projects/forms.pywafer_space/projects/migrations/0056_historicalproject_chip_on_board_and_more.pywafer_space/projects/migrations/0057_alter_manufacturabilitycheck_trigger_reason.pywafer_space/projects/models.pywafer_space/projects/tasks_checks.pywafer_space/projects/tests/constants.pywafer_space/projects/tests/test_forms.pywafer_space/projects/tests/test_models.pywafer_space/projects/tests/test_tasks.pywafer_space/projects/tests/test_views.pywafer_space/projects/views.pywafer_space/templates/projects/_file_display.htmlwafer_space/templates/projects/manufacturability_check_status.htmlwafer_space/templates/projects/project_detail.htmlwafer_space/templates/projects/project_form.html
Comment on lines
+208
to
+225
| def test_detail_shows_cob_change_badge_in_check_history(self): | ||
| """Detail page shows CoB Change badge for COB_CHANGE trigger in history.""" | ||
| project_file = ProjectFileFactory(project=self.project) | ||
| self.project.submitted_file = project_file | ||
| self.project.save() | ||
| ManufacturabilityCheckFactory( | ||
| project=self.project, | ||
| project_file=project_file, | ||
| status=ManufacturabilityCheck.Status.FINISHED, | ||
| trigger_reason=ManufacturabilityCheck.TriggerReason.COB_CHANGE, | ||
| ) | ||
| self.client.login(username="testuser", password=TEST_PASSWORD) | ||
| url = reverse("projects:detail", kwargs={"pk": self.project.pk}) | ||
|
|
||
| response = self.client.get(url) | ||
|
|
||
| assert response.status_code == HTTP_OK | ||
| assert "CoB Change" in response.content.decode() |
There was a problem hiding this comment.
test_detail_shows_cob_change_badge_in_check_history does not guarantee history-path coverage.
On Lines 213-218, only one COB_CHANGE check is created. In _file_display.html, history badges are rendered only for non-first entries, so this can pass without validating the history mapping branch. Add a newer check so COB_CHANGE is definitely rendered as history.
Suggested test-shape adjustment
ManufacturabilityCheckFactory(
project=self.project,
project_file=project_file,
status=ManufacturabilityCheck.Status.FINISHED,
trigger_reason=ManufacturabilityCheck.TriggerReason.COB_CHANGE,
)
+ # Create a newer/latest check so COB_CHANGE is shown in history, not current.
+ ManufacturabilityCheckFactory(
+ project=self.project,
+ project_file=project_file,
+ status=ManufacturabilityCheck.Status.FINISHED,
+ trigger_reason=ManufacturabilityCheck.TriggerReason.RETRY,
+ )🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@wafer_space/projects/tests/test_views.py` around lines 208 - 225, The test
test_detail_shows_cob_change_badge_in_check_history currently only creates one
ManufacturabilityCheck (via ManufacturabilityCheckFactory) so the COB_CHANGE
entry may be treated as the first entry and not rendered by the history branch
in _file_display.html; update the test to create an earlier (older) check for
the same project_file (e.g., a finished check with a different trigger_reason or
earlier timestamp) before creating the COB_CHANGE check so that the COB_CHANGE
check is a non-first history entry and the history-path rendering is exercised.
Comment on lines
+499
to
+512
| def test_unchanged_cob_creates_no_check(self): | ||
| """Submitting the form with CoB unchanged creates no new check.""" | ||
| check = self._make_submitted_check(ManufacturabilityCheck.Status.FINISHED) | ||
| self.client.login(username="testuser", password=TEST_PASSWORD) | ||
| url = reverse("projects:update", kwargs={"pk": self.project.pk}) | ||
|
|
||
| self.client.post(url, self._cob_form_data(chip_on_board=False)) | ||
|
|
||
| assert ( | ||
| ManufacturabilityCheck.objects.filter( | ||
| project_file=check.project_file | ||
| ).count() | ||
| == 1 | ||
| ) |
There was a problem hiding this comment.
test_unchanged_cob_creates_no_check can false-pass when form POST is invalid.
On Line 505, the response is ignored; if validation fails, check count can stay 1 and the test still passes. Assert the response status and project flag to ensure the request succeeded and the “unchanged” path was actually exercised.
Suggested hardening
- self.client.post(url, self._cob_form_data(chip_on_board=False))
+ response = self.client.post(url, self._cob_form_data(chip_on_board=False))
+ assert response.status_code == HTTP_FOUND
+ self.project.refresh_from_db()
+ assert self.project.chip_on_board is False
assert (
ManufacturabilityCheck.objects.filter(
project_file=check.project_file
).count()
== 1
)📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| def test_unchanged_cob_creates_no_check(self): | |
| """Submitting the form with CoB unchanged creates no new check.""" | |
| check = self._make_submitted_check(ManufacturabilityCheck.Status.FINISHED) | |
| self.client.login(username="testuser", password=TEST_PASSWORD) | |
| url = reverse("projects:update", kwargs={"pk": self.project.pk}) | |
| self.client.post(url, self._cob_form_data(chip_on_board=False)) | |
| assert ( | |
| ManufacturabilityCheck.objects.filter( | |
| project_file=check.project_file | |
| ).count() | |
| == 1 | |
| ) | |
| def test_unchanged_cob_creates_no_check(self): | |
| """Submitting the form with CoB unchanged creates no new check.""" | |
| check = self._make_submitted_check(ManufacturabilityCheck.Status.FINISHED) | |
| self.client.login(username="testuser", password=TEST_PASSWORD) | |
| url = reverse("projects:update", kwargs={"pk": self.project.pk}) | |
| response = self.client.post(url, self._cob_form_data(chip_on_board=False)) | |
| assert response.status_code == 302 | |
| self.project.refresh_from_db() | |
| assert self.project.chip_on_board is False | |
| assert ( | |
| ManufacturabilityCheck.objects.filter( | |
| project_file=check.project_file | |
| ).count() | |
| == 1 | |
| ) |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@wafer_space/projects/tests/test_views.py` around lines 499 - 512, The test
test_unchanged_cob_creates_no_check currently ignores the POST response so it
can false-pass on form validation errors; after calling self.client.post(url,
self._cob_form_data(chip_on_board=False)) capture the response, assert its
status_code indicates success (e.g., 302 for redirect after successful POST or
200 if no redirect), then refresh the related object (use the project_file
returned by self._make_submitted_check or reload it from the DB) and assert its
chip_on_board/project flag remains unchanged, and finally assert
ManufacturabilityCheck.objects.filter(project_file=check.project_file).count()
== 1 to ensure no new check was created.
Addresses a CodeRabbit finding on PR #265: "latest" and is_cancellable were read off an unlocked in-memory instance, so a concurrent transition could (a) clobber a freshly FINISHED check back to CANCELLING from a stale in-memory status, or (b) let two simultaneous CoB toggles each create a COB_CHANGE check. Wrap the body in transaction.atomic() and re-read the row with select_for_update() before any decision, so the choice is made against committed state and concurrent toggles serialize on the row lock. The new test reproduces the FINISHED-clobber deterministically (no threads): it updates the row out-of-band, then asserts the locked re-read leaves the committed FINISHED status untouched. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M1FsYafhYvbwqWVCkGDPYJ
Addresses a CodeRabbit finding on PR #265: create_check_cob_change() can raise ValueError when the source check is no longer its file's latest. Because it runs after super().form_valid() has already saved the project, and ATOMIC_REQUESTS wraps the whole request in one transaction, the raise rolled the request back into a 500 and discarded the user's valid CoB edit. Catch the recoverable ValueError, log a warning, and let the request finish normally. The model's transaction.atomic() block means the inner savepoint rolls back cleanly, leaving the committed project save intact; the newer check that superseded this one picks up the new flag on its own. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M1FsYafhYvbwqWVCkGDPYJ
The project detail page showed non-Chip-on-Board projects as "Standard" packaging. The correct term for an unpackaged die is "Bare Die", so rename the badge label and update the detail-view test accordingly. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01M1FsYafhYvbwqWVCkGDPYJ
4 tasks
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Closes #259.
Lets users request Chip-on-Board (CoB) packaging for a project. When set, the
manufacturability precheck runs with the precheck image's
--cobflag (the realargparse flag — the issue originally said
--chip-on-board), enabling the extraCoB compatibility checks. Toggling the option re-runs the check.
Design spec:
docs/superpowers/specs/2026-06-08-chip-on-board-packaging-design.mdImplementation plan:
docs/superpowers/plans/2026-06-10-chip-on-board-packaging.mdWhat changed
Project.chip_on_board— editable boolean (a USER_FIELD, not core-immutable); migrations0056(+HistoricalProjectmirror) and0057(newTriggerReasonchoice).TriggerReason.COB_CHANGE— new trigger reason for re-checks caused by a toggle.ManufacturabilityCheck.create_check_cob_change()— modelled oncreate_check_drc_update(), but cancels an in-progress check itself viamark_cancelling(reason="Chip-on-Board option changed")before creating the new PENDING check. This guarantees a superseded check can never FINISH with a result computed from the old CoB setting (ANALYZING→FINISHED is legal; CANCELLING permits only →CANCELLED). Pure ORM — no new Celery tasks or schedules; the existingchecks_pending/checks_cancellingpollers do the async work.do_starting— appends--cobto the precheck Docker command whencheck.project.chip_on_boardis set, read live from the project exactly like--slot/--id.ProjectUpdateView.form_validcreates the re-check when the flag changes (draft projects just persist the flag); Packaging badge on the project detail page;cob_changebranches added to all four trigger-reason badge chains (_file_display.htmlpreviously rendered no badge for unknown reasons).Relationship to #261 / #262
The design assumed PR #262 (
checks_cleanup_superseded) would land first, but does not depend on it: the explicit cancel increate_check_cob_change()is the mechanism; the scheduled superseded-cleanup is only defence-in-depth once #262 merges.Known gap (deliberate, needs a follow-up issue)
Django admin can edit
chip_on_board(ProjectAdminexposes all fields) without triggering the cancel+recreate logic, so an admin toggle during an in-flight check leaves it running with the old setting. The spec scoped the toggle hook to the project edit view; recommend a fast-follow to add asave_modelhook (or make the field admin-readonly) and addchip_on_boardto adminlist_display/list_filter.Test plan
create_check_cob_changebehaviors incl. cancel-on-RUNNING and non-latest guard, exact--cobcommand position, form presence/editability, view toggle/draft/unchanged paths, detail + check-history badges)make check-allclean (ruff, mypy, djlint)makemigrations --check— no missing migrations🤖 Generated with Claude Code
Summary by CodeRabbit