Plan 034: the Speckit plan for release 1 (standalone operation) of add-neutral-product-standalone-operability

[brettheap]

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

What this is

This PR is the Speckit plan for release 1, "standalone operation"
(phases 1–3), of the ratified OpenSpec change
add-neutral-product-standalone-operability (#1144). It encodes Brett Heap's
answers to the eleven questions that blocked phase 1.

It adds planning only: no code, no pin, and no OpenSpec change. It edits no
file of #1144. Besides the feature directory, it changes one line: an entry in
the README's Documentation index that links the feature's seven documents.

Its authority comes from two comments on #656:

• The ratification word, comment 5815412869: "ratify File add-neutral-product-standalone-operability: open the BUILD arc, recording ten rulings #1144, land the
  follow-ons, (a) on C1–C5".
• The answers, comment 5817152735 (2026-09-24T15:31:46Z): "(a) on all
  eleven, (d) on R1Q6".

The ratification record landed as #1151 → cd494e4c. This branch is cut from
it, and it merges main in at 1d14fee6.

The feature is specs/034-opendox-standalone-operation/:

file	what it holds
spec.md	User stories US1 (it runs), US2 (useful alone), US3 (installs) and US4 (the governed host unchanged). FR-001–FR-012, each mapped to a #1144 requirement and its boxes. § Clarifications, with the eleven answers as Q: … → A: …. The term-by-term definition of the AT-R1 acceptance test.
plan.md	Technical context. Constitution Check I–VII, where the one recorded deviation is now resolved for phase 1. The five target repositories. Where each box closes. The dependency graph and the single-writer surfaces. The pin and landing order (9.5). The trailer and Rule 6 rules. § Ruled answers and § Ruling needed (RN-1). In-flight overlaps.
research.md	Every measurement, with the command that reproduces it (W=$(mktemp -d); no host paths). Eight non-normative corrections to #1144 (R16).
tasks.md	87 tasks, T001–T098. Each task carries its repository, the boxes it realizes, its falsifier, and its Blocked by:, Ruled: and After: lines. T007 records the ruled amendments in #1144, and T008 raises the doc_health direction arc. T009 and T069 open phases 2 and 3. § Ruled amendments. Box accounting (69 of 124). The phase-1 writer slices, in parallel groups G1–G6.
quickstart.md	The AT-R1 procedure: the HTTP half (T095) and the browser half (T096).
clarify-questions.md	R1Q1–R1Q23: 11 ANSWERED, 12 OPEN
checklists/requirements.md	The spec-quality checklist. Its two open items are the constitution's own gates.

The answers (#656 5817152735), and where they go

question	answer	carried out by	#1144 amendment (T007)
R1Q22	(a) the manifest records arrival only	every carved-file task; T038 corrects runtime/cli.py:17-19	an 11.1 addendum (batch A)
R1Q1	(a) handler-contribution facet	T010, T011, T045; T084 and T086 in phase 3	2.2 and design.md § D4 addenda (A)
R1Q2	(a) named composition tests join 11.1's surfaces	T045, T094 (test_extension_point_parity.py)	11.1 and F11.1 widened (A)
R1Q3	(a), with (i) and (ii)	T016, T022	F3.1 line 2 and 3.2 (A); (ii) is RN-1
R1Q4	(a) openDox's own verbs; the host asserts	T015, T038, T046	none
R1Q5	(a) RuntimeSubcommand via SUBCOMMAND_EXTENSIONS, with the opendox-runtime alias	T038, T042	a 10.1 addendum (A)
R1Q6	(d) a declared exclusion with its count and its reason	T041, T043, T044; T008 (the direction arc)	F9.1 for openXdox-code, and 9.4 (B)
R1Q7	(a) a reviewed respelling-only allow-list	T043, T059, T086	F5.2 and 12.5's falsifier (C)
R1Q8	(a) tests_runtime/ included, with a DB service in the required job	T036	none (F9.1 unchanged)
R1Q9	(a) list_documents in phase 1	T025, T046	4.3's "four" (A)
R1Q20	(a) bookkeeping carries no trailer	T091, and this PR	an 11.0 addendum (A)

Each T007 batch is its own bookkeeping PR. Each lands under a Rule 6 window,
with no Arc: trailer, before the checkpoint that runs the amended falsifier.

RULING NEEDED

RN-1: requirement 3's fourth scenario, against R1Q3 (ii). The scenario
reads "the registered profile replaces the default for that process",
unconditionally. As ruled, (ii) refuses a host registration made after a parser
or server was built from the default. The recommendation is (a): the scenario's
WHEN gains "before the product's parser or server has been built from the
default", plus a scenario for the refusal. RN-1 holds no task, because the
after-build refusal is today's AlreadyRegistered (plan.md § "Ruling needed").

Still open (phases 2–3, and process)

• R1Q10–R1Q19 and R1Q21. Until they are answered, phases 2 and 3 are
  PROVISIONAL, and they authorize no implementation.
• R1Q23 is new, raised by R1Q6 (d). Four of 5.4a's six generator suites
  need doc_health, and F5.2 does not provide it. R1Q23 asks how F5.2 runs.
  It blocks T059 and T063, not phase 1.
• A watch item for the archive: under R1Q6 (d), requirement 9 stays an OPEN
  EXTRACTION for openXdox-code until the direction arc lands.

Phase-1 fan-out (tasks.md § "Phase 1 writer slices")

Before any slice runs, the holder runs T003 (ARC_BASE) and T005 (the
re-measure), then T006 (round 1a's analyze), and claims the slice (T002). The
groups then run in order:

• G1: P1-A (route seam), P1-B (default profile), P1-C (home-corpus seam)
  and P1-E (openxFactory reaches).
• G2: P1-D (authoring, default adapter) and P1-H (console script).
• G3: P1-F (sweep and nine-file repair), then P1-G (whole suite in CI).
• G4: P1-R (the openDox root pin).
• G5: P1-I (the openXdox pin and residue), then P1-J (openXdox green
  alone).
• G6: P1-K (the consumer pins and host wiring), then P1-L (T017, T018),
  then the T049 checkpoint. T049 also waits on T007's batches A and B, and on
  RN-1.

A slice is one writer and one claim in one repository. Its tasks land in
their After: order, one PR each unless a Lands with: line joins them.

Each slice's repository, files, dependencies, falsifier and model (Opus or
Sonnet) are in the table in tasks.md.

Copilot, round 1 (review 5306611493 at 437ca764)

All five threads were answered and resolved at 85d6de45:

• T031/T036 cycle. T031 is now After: T035, and it co-lands in T036's
  PR.
• T040/T047 cycle, and the same in phases 2 and 3. Each phase's pin
  advance is split into a root-pin task (T039, T062, T087) and a
  consumer-pin task (T047, T064, T094).
• SC-006. It now reads 65 of 69 at release-1 close.
• T003. It is no longer blocked, because R1Q20 is ruled.
• The checklist. Its repository item is narrowed to realization tasks.

Copilot, round 2 (review 5306959435 at 85d6de45)

All five threads were answered and resolved at d3a33165:

• T084 after T073. This is serve.py's single-writer order. The same
  audit also put T016 after T012 and T038 after T022. All six single-writer
  chains in plan.md are now ordered in the After: graph.
• The phase-2 graph shows T050 → T051. T025 and T051 lose [P].
• The quickstart's prerequisites name R1Q12, and so does T095's
  Blocked by: line.
• The quickstart's step 1 now asserts that there is no identity broker and
  that no database listens. Step 2 asserts that neither repository holds a
  model binding.
• R0 checks out all seven repositories at the commits its table names.

Copilot, round 3 (review 5307043972 at d3a33165)

Both inline threads were answered and resolved at 701a4a1a, and the three
"previously missed" items were answered in a PR comment:

• The constitution's workflow gate. Phases 2 and 3 are now PROVISIONAL.
  They authorize no implementation until each is re-planned on its answers
  and re-analyzed; since round 5 that is done by their round tasks, T009 and
  T069. Phase 1, the only phase the plan authorizes, has every material
  ambiguity resolved.
• spec.md's catalog edge case now separates the behaviour before T085
  from the behaviour after it.
• The acceptance graph is serial: T095 → T096 → T097.
• T017 and T093 run after each phase's openxFactory landing.
• T081 runs after T085, whose validators serve its catalog route.

Copilot, round 4 (review 5307117274 at 701a4a1a)

Both inline threads were answered and resolved at 0a5acd7c, and the three
"previously missed" items were answered in a PR comment:

• RN-1 against phase 1's close. RN-1 now holds T049. Requirement 3 is not
  reported as realized until RN-1 is ruled and T016 matches the ruling.
• T076 after T087. The README documents the command the root pins.
• T034 after T032, T047 after T007's batch A, and T059 after T007's batch
  C. The same audit ordered every arrow in the plan's graph against the
  After: graph.

Copilot, round 5 (review 5307230690 at 0a5acd7c)

Both inline threads were answered and resolved at 0ad046ba, and the two
"previously missed" items were answered in a PR comment:

• T006 waited on the open-ended T004. T004 is now round 1a, and it is DONE.
  T006 is round 1a's analyze, after T003, T004 and T005. Phases 2 and 3 open
  with their own round tasks, T009 and T069.
• The quickstart's readiness probe. It first checks that the port is free,
  and on every poll it checks that the launched PID is alive.
• Constitution row V. It records main's baseline and the accepted
  add-chain-attestation disposition, and notes that the enforced
  openspec-cli-pin gate exits 0.
• T053's trailer policy. openDox-spec is a sixth repository of the arc
  only if T053 applies, and T091 then covers it.
• The same audit:
  • co-landings now use a Lands with: line;
  • T026 loses [P];
  • serve.py's single-writer chain starts at T010.

Copilot, round 6 (review 5307694574 at 0ad046ba)

Its one thread was answered and resolved at 2a57501c:

• FR-006 now carries requirement 9's own clause for a check that runs
  less than the whole suite. openXdox-code declares its doc_health
  exclusion with a count and a reason, and reports it as an open extraction.
• Every other whole-suite claim carries the same qualifier.
• 9.2's close moves to phase 3, where T086 takes the ratchet to (0, 0).

Copilot, round 7 (review 5307775377 at 2a57501c: Findings: None)

Its three "previously missed" items were answered in a PR comment and are
addressed at abb2344e:

• T003 records openDox-spec's ARC_BASE unconditionally.
• T093 is the interim F11.1 procedure. Each phase runs it as its own
  task (T018, T065 and T098), and each checkpoint depends on its run.
• Constitution row IV records the README-index reading that every
  earlier Speckit feature follows.

Copilot, round 8 (review 5307865432 at abb2344e)

Its one thread was answered and resolved at 0a859498:

• The README document index (Principle IV). The README's Documentation
  section now carries an entry for feature 034, beside the openDox carve
  documents, and it links all seven documents. The entry sits outside the
  OpenSpec Records block, and Constitution row IV records it.

Copilot, round 9 (review 5307941798 at 0a859498: Findings: None)

Its two "previously missed" items were answered in a PR comment and are
addressed at 1522c4c7:

• T095 against T036's PostgreSQL service. AT-R1's HTTP half is now a
  harness, acceptance/at_r1_http.py, run by its own acceptance job with no
  database service. It is not a pytest module, so it changes neither
  openDox-code's testpaths (which T036 sets in phase 1) nor F9.1, and
  FR-006's "no exclusion" still holds.
• T072's falsifier no longer names the procfs path, and the quickstart's
  null-device redirects are gone. The host-path scan now covers the system
  roots as well.

2e4f3315 then merges main in at 1d14fee6.

Copilot, round 10 (review 5308148911 at 2e4f3315)

Its two threads were answered and resolved at 34b7b2ca:

• The harness and T036. The plan and T095 now say the harness changes
  neither testpaths, which T036 sets in phase 1, nor F9.1.
• T073 after T055. T073 reached T055 only through phase 2's checkpoint,
  and it now names T055 directly. The seven other single-writer pairs that
  relied on a transitive path are direct edges as well, so all 19 consecutive
  pairs of the six chains are direct.

Copilot, round 11 (review 5308305662 at 34b7b2ca: Findings: None)

Its one "previously missed" item was answered in a PR comment and is addressed
at bc1bd20b:

• T095 after T076. AT-R1 runs the root README's documented command, so
  T095 now waits for T076, the task that writes that README. So does T089,
  which leaves no phase-3 task outside phase 3's checkpoint.
• The same check on every phase. T049 now names T017, as the P1-L row
  already said. T017's After: line no longer lists its phase-2 and phase-3
  re-runs.

Copilot, round 12 (review 5308403738 at bc1bd20b: Findings: None)

Its one "previously missed" item was answered in a PR comment and is addressed
at a86ced35:

• T096 had no repository tag and no falsifier. It now carries [oxF],
  where its evidence lands, and names the oracle's verdict. The same check over
  every task found eight more gaps (T074, T077, T083, T090–T093), and each now
  names its repository and its falsifier. The checklist item names these
  classes.

Copilot, round 13 (review 5308511992 at a86ced35)

Its one thread was answered and resolved at 49eb53ee:

• macOS could not run F13.1. The plan's Target Platform now names Linux
  alone. 13.1 checks the "no TCP port" invariant in the Linux kernel's socket
  table, and no ratified falsifier covers macOS. Adding macOS would need a
  Darwin leg for 13.1's check and F13.1, which is an amendment to File add-neutral-product-standalone-operability: open the BUILD arc, recording ten rulings #1144 and so
  needs a ruling.
• From the same pass: T036 now comes after T038, because both edit
  openDox-code's pyproject.toml, and T046's range skips two undefined ids.

Copilot, round 14 (review 5308583172 at 49eb53ee)

Its one thread was answered and resolved at e927069e:

• T092's note versus "no field changes". T092 and its falsifier now state
  F11.1's own rule. With every edits[].note removed, the manifest is
  unchanged, and an existing note may only be extended.
• From the same pass: three checklist claims are now exact (FR-011 and FR-012,
  SC-006 and SC-007, and the edge cases).

Copilot, round 15 (review 5308662086 at e927069e: Findings: None)

Its one "previously missed" item was answered in a PR comment and is addressed
at 9b86b920:

• R4 and R10 ran ci_coverage.py with the host's python3. They now use
  the od and ox venvs, and so do R5's, R6's and R9's runs. From the same
  pass, T096 now also records T095's acceptance job for SC-004, and spec.md's
  Status header reads Draft, as the sibling features' headers do.

Copilot, round 16 (review 5308754464 at 9b86b920)

Its one thread and its one "previously missed" item were answered and are
addressed at 4d3c605b:

• US4's note condition now admits a note added where an entry had none,
  as T092 and F11.1 do.
• T053's condition is now "R1Q11 (a) or R1Q12 (b)" in the plan's graph,
  its phase-2 lanes and Constitution row VI, as T053 itself says.

Copilot, round 17 (review 5308830806 at 4d3c605b)

Findings: None, and nothing listed as previously missed. All 23 inline
threads are resolved.

The adversarial sweep (at 4d3c605b), folded in at 1cb01e4c

Round 17 was clean, and every check was green at 4d3c605b. Each earlier round
had still found one or two new cross-document gaps, so a read-only Opus
reviewer swept all seven files, #1144 and the measured code. It reported 28
MUST and 23 NICE findings. Every MUST was verified against the text, the packet
or the code, and fixed; the exact NICEs were fixed too. The PR comment at
1cb01e4c lists them. The phase-1 ones were:

• T036 gives validate the .[runtime,test] install as well as its PostgreSQL
  service. Without it, tests_runtime/conftest.py fails every runtime case
  under CI.
• T043 waits for C3's PR 2, because tests/test_snapshot.py fails on the
  validator lookup, not on doc_health.
• openxFactory's test_serve_column_split.py is the second named composition
  test.
• A slice's tasks land in After: order, and T032 is the join of every lane.
• Batch D would carry RN-1's scenario text, only on Brett's ruling.

Copilot, rounds 18–24

round	review	head	result	addressed at
18	5309539162	9f799f20	1 thread: show T026 → T025 in the graph	902da25b: the single-writer table lists every phase-1 shared file (workbench.py, openDox-code pyproject.toml, tests/test_authoring_seam.py)
19	5309640972	902da25b	1 thread (how T035's six modules move) and 1 previously missed (Status: headers)	224046c5: each module is rewritten or re-landed by T042, and all seven files carry Status: draft
20	5309742955	224046c5	1 thread: T049 must accept the openxFactory destination	53a85ed9
21	5310503144	53a85ed9	1 thread (T006's feature context) and 1 previously missed (row V's base)	09e81dc3: SPECIFY_FEATURE_DIRECTORY and the prerequisite check; the gate is re-run on main at 138a4722
22	5310581314	09e81dc3	1 thread (the T043/T061 contingency cycle) and 2 previously missed (the outline-model route, openXdox-spec)	02b1aa30: the contingency is caught by T005 before phase 1; openXdox-spec is a conditional seventh repository
23	5310647715	02b1aa30	Findings: None, nothing previously missed	243e460f: a self-check found that T044 must re-pin openXdox's floors when a file joins the exclusion
24	5310714316	243e460f	Findings: None, nothing previously missed	— (the head)

Validation (head 243e460f)

gate	result
OPENSPEC_TELEMETRY=0 openspec validate --all --strict	Totals: 110 passed, 1 failed (111 items) at the head, and the same on main itself at 138a4722, which the branch merges. The one failure is add-chain-attestation's ratified disposition.
python3 scripts/validate-openspec-cli-pin.py --all	exit 0; every target validated --strict with 0 UNDISPOSITIONED failures
the After: graph in tasks.md, checked by DFS	no cycle; the same check reports the three pin-advance cycles at 437ca764
plan.md's nine single-writer chains, each consecutive pair checked against the After: graph	0 gaps; all 22 pairs are direct After: edges
the 50 arrows in plan.md's graph and tasks.md's Dependencies, each checked against the After: graph	0 mismatches (6 at d3a33165)
every realization task outside Phase 0 names its repository and its falsifier; no Lands with: pair names each other on After:	0 gaps
every task of each phase before its checkpoint (T049, T063, T089), a Lands with: pair counting as one landing	0 gaps; T048, the holder's F4 re-measure, follows T049 by design
phase gates: every phase-1 task after T006, every phase-2 task after T009, every phase-3 task after T069	0 missing
root pins T039, T062, T087: every openDox-code task of their phase an ancestor (a Lands with: pair counts as one landing)	0 missing
checkpoints T049, T063, T089 after their interim F11.1 runs T018, T065, T098	all three
question headers against the tasks' Blocked by:/Ruled: lines	23 headers, 0 mismatches
box table	69 release-1 boxes, each naming defined tasks
host-absolute path scan (home, scratch and system roots) and closing-keyword scan (files and commit messages)	0 hits

Process

• No Rule 6 window is needed. This PR touches no openspec/changes/
  directory and not the README's OpenSpec Records block: its one README line
  is in the Documentation index. T007's batches and T097's ticks will each
  need a window.
• No Arc: trailer. This PR is bookkeeping, and R1Q20 (a) is now ruled.
• Sibling reads (Rule 1).
  • openDox + openXdox: two open-source layers, an installable app with users, projects and a database; domain descendants pin openXdox (Brett's ruling 2026-09-04) #656: no claim on the release-1 planning object.
  • gh pr list -R opensoft/openxFactory --state all --search 034-opendox:
    only File add-neutral-product-standalone-operability: open the BUILD arc, recording ten rulings #1144.
  • specs/034-*: absent on main.
• Lane 4's in-flight acts have all landed (plan.md § "In-flight overlaps").
  • C1 landed as Correct the stale 124-of-143 current-corpus count to 129 of 183 (D-13, RULED 5815412869 C1) #1152 → 9afea8d7.
  • C3's PR 1 landed as Declare the § 8.9 snapshot-validator residue window in the carve manifest (Q-L1 form; RULED 5815412869 C3) #1153 → 1d14fee6, and its PR 2 as openXdox-code#28 →
    e28930bf. C3 overlaps 7.3 (T061) and the openXdox pin (T047). T061's
    After: condition on PR 2 is now met.
  • C3's openxFactory openXdox pin bump landed as Consume the confined snapshot-validator locator: openXdox pin 2f3f857d → 069fe471, with its two consumers (RULED 5815412869 C3) #1157 → 1edbb3dd, and C4's
    runbook as Write the openXdox pin resync runbook the pin said was owed (D-16, RULED 5815412869 C4) #1154 → 138a4722. T090 follows the runbook.
• Persisted tools. Everything is in opensoft/brett-wip, under
  handoffs/xFactory/attachments/lane-opendox/r1plan/tools/:
  • the four measurement scripts and the shim, at 011b1586;
  • the plan-consistency checks (qmap.py, qcheck.py, depcheck.py,
    chaincheck.py, chaindirect.py, arrowcheck.py, phasecover.py and
    sharedfiles.py), to re-run after every re-plan.

🤖 Generated with Claude Code

[Copilot]

Copilot review overview

🟡 Changes recommended

Critical task-ordering cycles and several dependency, validation, and checklist inconsistencies remain unresolved.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 2 · 2 · 1

Open (5)

• T031 and T036 have a circular dependency · New
• Phase-1 ordering contains a T040/T047 dependency cycle · New
• Release-1 success criterion reports an incorrect completed-box count · New
• T003 is incorrectly listed as unblocked despite R1Q20 · New
• Checklist incorrectly requires repositories on holder and checkpoint tasks · New

What changed in this PR

Planning-only PR for release 1 of openDox standalone operation.

Changes:

• Defines requirements, acceptance criteria, and AT-R1.
• Adds research, architecture, dependencies, pinning, and landing plans.
• Adds 97 phased tasks, clarification questions, and readiness checklist.

File	Summary
specs/​034-opendox-standalone-operation/​tasks.md	Phased realization tasks, dependencies, box accounting, and execution order
specs/​034-opendox-standalone-operation/​spec.md	Feature requirements, scenarios, and acceptance criteria
specs/​034-opendox-standalone-operation/​research.md	Reproducible measurements and analysis
specs/​034-opendox-standalone-operation/​quickstart.md	AT-R1 acceptance procedure
specs/​034-opendox-standalone-operation/​plan.md	Technical approach, phases, dependencies, and landing strategy
specs/​034-opendox-standalone-operation/​clarify-questions.md	Open implementation decisions
specs/​034-opendox-standalone-operation/​checklists/​requirements.md	Specification readiness checklist

---

💡 Configure MCP servers for context-aware, tailored reviews. Learn more in the docs.

[Copilot]

Copilot review overview

🟡 Changes recommended

Unresolved dependency, prerequisite, reproducibility, and acceptance-procedure inconsistencies must be addressed.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1 · 4

Open (5)

• Make T084 depend on T073 for serve.py writes · New
• Add T050 → T051 dependency to the Phase 2 graph · New
• Include R1Q12 in quickstart prerequisites · New
• Add explicit absence checks for services and bindings · New
• Pin all seven R0 repositories for reproducible runs · New

Resolved since last review (5)

• Phase-1 ordering contains a T040/T047 dependency cycle
• T031 and T036 have a circular dependency
• T003 is incorrectly listed as unblocked despite R1Q20
• Release-1 success criterion reports an incorrect completed-box count
• Checklist incorrectly requires repositories on holder and checkpoint tasks

[Copilot]

Copilot review overview

🟡 Changes recommended

Unresolved critical and moderate issues affect conformance, scope, reproducibility, and task sequencing.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 2

Open (2)

• Unresolved ambiguities violate mandatory planning gate · New
• Edge case contradicts standalone validator contract · New

Resolved since last review (5)

• Make T084 depend on T073 for serve.py writes
• Pin all seven R0 repositories for reproducible runs
• Add explicit absence checks for services and bindings
• Include R1Q12 in quickstart prerequisites
• Add T050 → T051 dependency to the Phase 2 graph

Previously missed (3)

In code that hasn't changed since last review

Phase graph conflicts with T096 dependency on T095

specs/​034-opendox-standalone-operation/​plan.md:195

The phase graph says T095 and T096 run in parallel, but tasks.md makes T096 explicitly After: T095. The plan and the executable task schedule therefore disagree about the acceptance order; if the task dependency is intentional, the graph should show the same serial chain.

P1-L lacks dependency on P1-K landing

specs/​034-opendox-standalone-operation/​tasks.md:219

The phase-1 slice table says P1-L runs after P1-K, but this task's After: — allows it to be scheduled before the T047 pin/host landing whose F11.1 output it is meant to record. Add the phase-1 landing dependency so the task-level DAG matches the slice ordering.

T081 can land before required T085 validator

specs/​034-opendox-standalone-operation/​tasks.md:756

T081 says the served catalog route is part of this task but requires T085, whose validators are also blocked by R1Q12; its After: T063 lets the task land before that prerequisite. Add T085 to the dependency, remove the [P] marker, and update the phase-3 graph/slice so the no-model route is not accepted before its validator default exists.

[brettheap]

Answering the three "Previously missed" items in Copilot review 5307043972 (at d3a33165). They carry no inline thread, so the answers are here. All three are fixed in 701a4a1a.

• The phase graph conflicts with T096's dependency on T095 (plan.md:195). The acceptance line now reads T095 (HTTP, CI) → T096 (browser) → T097, which matches tasks.md's After: T095.
• P1-L lacks a dependency on P1-K (tasks.md:219). T017 now runs After: T047, and it is re-run after T064 and T094. T093 now runs after each phase's openxFactory landing: T047, T064 and T094. T017 loses [P], and the phase-1 parallel summary puts T017 and T093 after T047.
• T081 can land before the T085 validator (tasks.md:756). T081 now runs After: T063, T085. Its Blocked by: line names R1Q12 for the served route, and it loses [P]. The phase-3 graph shows T085 → T081, and shows T084 after T073. The Dependencies section and the parallel summary say the same.

The two inline findings are answered in their threads, and both threads are resolved.

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

[Copilot]

Copilot review overview

🟡 Changes recommended

Unresolved moderate planning inconsistencies affect requirements, phase gates, and task dependencies.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 2

Open (2)

• Phase 1 claims completion despite unresolved requirement-3 behavior · New
• T076 lacks dependency on final code pin T087 · New

Resolved since last review (2)

• Edge case contradicts standalone validator contract
• Unresolved ambiguities violate mandatory planning gate

Previously missed (3)

In code that hasn't changed since last review

T034 lacks dependency on the T032 import inventory sweep

specs/​034-opendox-standalone-operation/​tasks.md:335

The execution graph explicitly orders T032 → T034, and this task re-derives the consumer-reach and boundary censuses that T032's sweep establishes. With only After: T011, T034 can be started before that sweep and can produce a result inconsistent with the planned import inventory. Add T032 to this dependency.

T047 lacks dependency on T007 batch A

specs/​034-opendox-standalone-operation/​tasks.md:498

T047 carries T045's parity-test edit, but the task can currently land before T007 batch A widens F11.1 to admit that named composition test. The resulting Arc landing would be rejected by the guard even though the writer-slice table declares batch A as a prerequisite; add that batch to T047's dependency.

T059 lacks dependency on the reviewed R1Q7 allow-list

specs/​034-opendox-standalone-operation/​tasks.md:625

T059 edits the protected generator suites while its own falsifier requires the R1Q7 allow-list recorded by T007 batch C, but it does not wait for that batch. Since T059 runs before T063 (the task that currently waits on batch C), the protected-suite edit can land without the required reviewed entry; add batch C to T059's prerequisites.

[brettheap]

Answering the three "Previously missed" items in Copilot review 5307117274 (at 701a4a1a). All three are fixed in 0a5acd7c.

• T034 lacks a dependency on T032's sweep (tasks.md:335). T034 now runs After: T032.
• T047 lacks a dependency on T007's batch A (tasks.md:498). T047 now runs After: T039, T044, T007 (batch A). T007 now names the tasks that wait on each batch: batch A before T047 (and so before T049), batch B before T049, and batch C before the first respelling edit and before T059.
• T059 lacks a dependency on the R1Q7 allow-list (tasks.md:625). T059 now runs After: T052, T055, T062, T007 (batch C).

The same audit ordered every arrow the plan draws. T022 runs after T021, T042 after T041, T053 and T061 after T049, and T075 and T085 after T063. A check of the 46 arrows in plan.md's graph and in tasks.md's Dependencies section against the After: graph reports 0 mismatches at 0a5acd7c. At d3a33165 the same check reported 6.

The two inline findings are answered in their threads, and both threads are resolved.

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

[Copilot]

Copilot review overview

🟡 Changes recommended

Critical acceptance-test and task-scheduling issues, along with unresolved governance and consistency findings, block approval.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 2

Open (2)

• Tie readiness probe to launched service · New
• Resolve T006 dependency on incomplete T004 · New

Resolved since last review (2)

• T076 lacks dependency on final code pin T087
• Phase 1 claims completion despite unresolved requirement-3 behavior

Previously missed (2)

In code that hasn't changed since last review

Resolve failing validation before marking Principle V pass

specs/​034-opendox-standalone-operation/​plan.md:87

The Constitution Check marks Principle V as PASS for this PR, but the validation result supplied for this branch is 110 passed, 1 failed; Principle V requires openspec validate --all --strict to pass (.specify/memory/constitution.md:77-80). If the pre-existing add-chain-attestation failure is intentionally tolerated, record that baseline disposition and why it is unaffected here; otherwise this row cannot be marked PASS.

Define trailer policy for conditional repository

specs/​034-opendox-standalone-operation/​tasks.md:890

T053 makes openDox-spec a possible realization repository, but this policy says every realization landing exists in only five repositories. If that conditional path is selected, the schema commit needs the same Arc/Lane trailer and landing rules, or the task must explicitly be classified outside this policy; otherwise the release cannot satisfy its own trailer contract.

[brettheap]

Answering the two "Previously missed" items in Copilot review 5307230690 (at 0a5acd7c). Both are fixed in 0ad046ba.

• Constitution row V was marked PASS while the strict validation reports a failure (plan.md:87). The row now reads "PASS for this PR, against main's recorded baseline", and it records that baseline:
  • This PR touches no openspec/ path.
  • openspec validate --all --strict gives 110 passed, 1 failed here, exactly as on main at cd494e4c.
  • The one failure is add-chain-attestation. It is an accepted disposition in contracts/openspec-cli-pin.yaml (ratified_by: 'Brett Heap, 2026-09-05, "take exit 2"'), and it retires when that change archives.
  • The gate the repository enforces, scripts/validate-openspec-cli-pin.py --all (the openspec-cli-pin check), exits 0 with 0 UNDISPOSITIONED failures.
• The trailer policy for a conditional repository (tasks.md:890). T091 now covers every repository the arc touches: the five, and openDox-spec when T053 applies, which is 11.0's own "in EVERY repository it touches". T053 states that its landings carry the Arc: and Lane: trailers and land by squash or merge. openDox-spec allows squash and merge and requires validate, and T003 records its ARC_BASE. plan.md and spec.md count openDox-spec as a sixth repository only if T053 applies.

The two inline findings are answered in their threads, and both threads are resolved.

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

[Copilot]

Copilot review overview

🟡 Changes recommended

Resolve the documented planning, dependency, acceptance, and procedure gaps before approval.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1

Open (1)

• FR-006 conflicts with the declared doc_health exclusion · New

Resolved since last review (2)

• Resolve T006 dependency on incomplete T004
• Tie readiness probe to launched service

[Copilot]

Copilot review overview

🟡 Changes recommended

Resolve the task dependency, conditional-path, checkpoint, and reproducibility inconsistencies before approval.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1

Open (1)

• Allow the openxFactory destination in T049 validation · New

Resolved since last review (1)

• Define ignored module relocation before openDox-code suite cleanup

[Copilot]

Copilot review overview

🟡 Changes recommended

Unresolved critical and moderate findings affect phase-1 execution, dependency ordering, and validation coverage.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1

Open (1)

• Establish feature context before phase gate · New

Resolved since last review (1)

• Allow the openxFactory destination in T049 validation

Previously missed (1)

In code that hasn't changed since last review

Validation uses incorrect base commit

specs/​034-opendox-standalone-operation/​plan.md:109

This validation claim uses main at 138a4722, but research R0 records the measured openxFactory baseline as dd2466ad, while this plan itself identifies 138a4722 as the C4 landing before that later main baseline. Reconcile the branch/measurement commit and rerun or quote the gate against the actual base; otherwise the PASS statement is not reproducible and may omit subsequent changes.

[brettheap]

Round 21: answer to review 5310503144's previously missed item. The item has no inline thread. It is fixed in 09e81dc3.

Constitution row V cited a base it had not measured. Two commits were being run together in that row:

• research.md's dd2466ad is the baseline the plan's measurements were taken at;
• this PR's validation gate runs on the branch.

The gate has now been run on main itself at 138a4722, the commit the branch last merged. It ran in a temporary detached worktree of this lane's own clone, which was removed afterwards:

• OPENSPEC_TELEMETRY=0 openspec validate --all --strict printed Totals: 110 passed, 1 failed (111 items), and the failure is add-chain-attestation.
• scripts/validate-openspec-cli-pin.py --all exited 0: 0 UNDISPOSITIONED failures.

The branch head gives the same result. main has moved on to 57af6927, but its newer commits touch no openspec/, specs/, README.md or .specify/ path, so the gate is unchanged there.

Row V now says all of this, and it keeps dd2466ad apart as the measurement baseline.

The feature-context thread on T006 is answered and resolved. No thread remains unresolved.

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

[Copilot]

Copilot review overview

🟡 Changes recommended

A critical dependency deadlock and several unresolved planning and validation gaps must be addressed.

Get a fresh assessment by requesting another Copilot review.

Review effort: Lite
Findings: 1

Open (1)

• Resolve the T049 and T061 recovery-path deadlock · New

Resolved since last review (1)

• Establish feature context before phase gate

Previously missed (2)

In code that hasn't changed since last review

Assign or remove the alternative test relocation path

specs/​034-opendox-standalone-operation/​tasks.md:390

The second permitted resolution for test_outline_model.py makes it an openxFactory composition test, but T034 is an openDox-code task and T047's relocation/checkpoint accounting only covers modules removed by T035 (not this file). If that branch is chosen, no task owns moving and landing the test or adding its named-set entry before the pin, so coverage is either dropped or F11.1 fails. Assign this alternative to T047 with the required ordering, or remove the alternative.

Account for openXdox-spec in conditional schema work

specs/​034-opendox-standalone-operation/​tasks.md:677

R1Q12(b) explicitly re-homes the snapshot schemas from openXdox-spec, so this conditional path touches both spec repositories. T053 assigns only openDox-spec and openDox, while the plan and T091 account for only openDox-spec; no task owns removing/updating/validating openXdox-spec or its landing metadata. Add an openXdox-spec slice and update the repository/landing accounting, or make T053 conditional only on the new-schema path.

[brettheap]

Round 22: answers to review 5310581314's two previously missed items. Neither has an inline thread, and both are fixed in 02b1aa30.

1. test_outline_model.py's composition-test route had no owner. It now follows the same route as test_session_harness.py:
   • T034's PR removes the case and lists it, with its destination, in its body.
   • A T007 batch adds its openxFactory path to F11.1's named set.
   • T047's openxFactory PR lands it; T047 now names both possible composition tests.
   • T049 checks that it arrived. Its check now covers every module or case that T034 or T035 removed, at the destination its PR's list gives.
   • The P1-K slice row names both files.
2. R1Q12 (b) also touches openXdox-spec. (b) re-homes ideation-dashboard-snapshot and -index from openXdox-spec, so openXdox-spec is now a conditional seventh repository, with a new tag, [oXs].
   • T053 says what changes there: openXdox-spec retires its own copies in favour of openDox-spec's, and the openXdox root's spec pin moves.
   • Its landings carry the trailers (T091), T003 records its base, and T009 plans the slice once the answer is known.
   • The Format's repository list, the plan's source-code listing and heading, spec.md's repository count and R1Q12's option (b) all count it.

The deadlock thread on T043 and T061 is answered and resolved. No thread remains unresolved.

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

[Copilot]

Copilot review overview

🔵 Needs a closer look

Unresolved dependency-ordering, validation-coverage, reproducibility, and metadata issues remain.

Review effort: Lite
Findings: None

Resolved since last review (1)

• Resolve the T049 and T061 recovery-path deadlock

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
0.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[Copilot]

Copilot review overview

🔵 Needs a closer look

Moderate sequencing and dependency issues remain, along with documentation corrections.

Review effort: Lite
Findings: None

[sourcery-ai]

Sorry @brettheap, your pull request is larger than the review limit of 150,000 diff characters

[brettheap]

READY #1155 @243e460f 2026-09-25T13:18:34Z

• Head 243e460f20d0f3e1732fb2e9c48cdb94dd1132ef, base main. The branch last merged main at 138a4722. Today's main, 57af6927, merges cleanly (git merge-tree), and its newer commits touch no openspec/, specs/, README.md or .specify/ path. The PR is MERGEABLE and no longer a draft.
• Checks at the head: every check run is green. pytest-suite passed (21:52:23Z to 22:12:08Z), as did openxdox-consumer-gate, openspec-cli-pin, merge-master-approval, lane-line (re-run after the body edit), SonarCloud, clearing-dispatch-gate, former-id-arrival-gate, openreposhape-pin, release-tag-gate, signed-execution-chain-gate and wallet-validation. Sourcery skipped its review because the diff is over its 150,000-character limit.
• Copilot at the exact head: review 5310714316 at 243e460f (2026-09-24T21:57:35Z) reports Findings: None, with nothing listed as previously missed. Round 23 at 02b1aa30 read the same.
• Threads: all 28 have a reply and are resolved. Every "previously missed" item from every round is answered in a PR comment.
• Process:
  • No Rule 6 window is needed. The PR touches no openspec/changes/ path and not the OpenSpec Records block; its one README line is in the Documentation index.
  • It carries no Arc: trailer, because it is bookkeeping (R1Q20 (a)).
  • It is not an archive, and no commit message or PR body carries a closing keyword.
  • RN-1 stays RULING NEEDED, and File add-neutral-product-standalone-operability: open the BUILD arc, recording ten rulings #1144 is not edited.
• Landing and fan-out: it lands through the holder's lander; this actor merges nothing. The phase-1 slice list (P1-A to P1-L, groups G1 to G6) is in tasks.md § "Phase 1 writer slices", and it is copied at this head into brett-wip handoffs/xFactory/attachments/lane-opendox/r1plan/PROGRESS.md.

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

[brettheap]

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

LANDING — lane openxfactory-4, session faabad8b-8c31-4958-abd5-b35bdc9cd282@Eagle, 2026-09-25T13:29:38Z, PR #1155 into opensoft/openxFactory main

[brettheap]

Lane: openxfactory-4 (openXfactory-4-openDox_extraction)

LANDED — lane openxfactory-4, 2026-09-25T13:30:02Z, PR #1155 → c415c3d (opensoft/openxFactory main; squash)