Repository navigation
Triage every open issue against the map's destination and produce a close-list #375
Description
Question
40 issues are open, once the seven ruled out while charting are closed and this map's own tickets are excluded. Half of the last week's intake was repo-scoped process work rather than product
work. Triage every open issue against this map's Destination and produce a close-list for the
maintainer to approve in one pass.
Per issue, one line: number, title, and either keep with the reason it serves the destination, or
close with the reason it sits past it. Do not close anything before approval. The seven already
ruled out while charting are done — do not re-litigate them.
Watch three traps:
- Priority does not decide this.
#332is p1 and still off the install-and-run path;#372is p2 and
breaks night-shift. #362and#372wait on the doctrine ticket, not on this one — deleting §4 may remove what they
describe.unic-pr-review(#275,#274,#316) is a separate live product. Leave it open and untouched.
Acceptance criteria
- Every open issue appears exactly once in the list, with a keep-or-close verdict and a one-line
reason. - The stream tickets
#313,#314,#315,#317,#344each carry a verdict, and closing a
stream is proposed only when its children are also closing. -
#275,#274and#316are marked keep, named as outside this map. - Nothing is closed until the maintainer approves the list.
- After approval, each closed issue carries a comment naming this map and the reopen path, and the
map's Out-of-scope section lists it.
Activity
orioltf commentedon Aug 14, 2026
Draft close-list — not approved, nothing closed
Produced while charting the map (2026-08-14), against the 40 open non-wayfinder issues at that moment.
The session that claims this ticket must re-check it against the tree at that time — issues move,
and this draft is a starting point, not a verdict. No issue was closed.
Close — 33
lib/andtest/go)slopcheckflag · #355validateConfig· #371artefact-install· #300 new test over prose/setuprewrite (#383)first installline · #302 scratch-Consumer clone —DXP-DesignSystemis the Consumer now/cleanup· #282 wayfinder as a Boxunic-ticket-specification, shipped and running inDXP-ProfileServicesKeep — 7
unic-pr-review— a separate live product, in use at the Consumer todayresolvedorclosed. On Azure DevOps with a client watching the board, a stale work-item state is a product defect, not processResult: the non-wayfinder backlog goes 40 → 7.
Four the maintainer decides explicitly
ready-for-agentlets an unattended run re-dispatch it;a killed Archon run can commit work and open no pull request. Both are real night-shift hazards,
and night shift is the speed mechanism. Decision 2 closes them because neither blocks the Consumer.
config-schema.mjsforcesthe defaults into comments anyway. It is real work either way.
Consumers. Closed here because Rewrite commands/setup.md from scratch as prose #383 and Prune the ADRs aggressively #384 make it wrong again; the rewrite belongs after them.
real evidence.
orioltf commentedon Aug 14, 2026
How to start this session
Produces: an approved close-list, applied, and the map's Out-of-scope section grown.
Run
/wayfinder 373 #375. It claims this ticket.Read the draft close-list comment above. It was written on 2026-08-14 and is explicitly not
approved. Re-check every number against today's tree — issues move.
Refresh the real list:
gh issue list --state open --limit 200 --json number,title,labels \ --jq '.[]|select([.labels[].name]|any(startswith("wayfinder:"))|not)|"#\(.number) \(.title)"'Present the corrected list to the maintainer. Close nothing before approval.
After approval, per issue: comment naming map #373
and the reopen path, then
gh issue close <n> --reason "not planned". feature(unic-ticket-specification): portable ticket-specification Archon workflow plugin #256 closes as completed,not as not-planned — it shipped.
Append every closed issue to the map's Out-of-scope section, one line and a link each.
Traps
unic-pr-review(bug(unic-pr-review): every ADO Writer az invoke uses wrong --resource/--api-version — all --post writes fail (--resource and --api-version combination is not correct) #275, bug(unic-pr-review): intent-check merger discards all verdicts when an item has anote— Intent Check silently renders as all-unaddressed #274, stream: unic-pr-review fixes #316, [unic-pr-review] Advisory Notice for Work Items mentioned in the PR title but not linked #253, [unic-pr-review] Optional non-authoritative ADO Fetcher debug dump #254, unic-pr-review: false-premise 🔴 critical — diff-only agents assert runtime behavior without tracing data flow #258) stays open. It is a separate live product.orioltf commentedon Aug 14, 2026
Correction to the draft close-list — #303
The draft puts #303 (
/archon-rolloutmust skip spec issues and wayfinder maps) under "Process work, frozen by decision 2". That reads wrong. It is dispatch safety, not tracker ergonomics: three kinds of issue carryready-for-agentand only one is dispatchable, so a sweep can send an agent at a spec issue described as user stories, with no acceptance criteria and no blocking edges.#298 already tried the documentation route across three files. #303 exists because that was judged insufficient — its own words: "Documentation asks a human to remember; this issue makes the command enforce it." PR #386 adds a qualifier to the state-roles table, which stops that table contradicting the paragraph below it and does nothing about the sweep.
Keep #303 open. It is not p0. Two facts hold the risk down: map #373's tickets carry
wayfinder:*labels and noready-for-agent, so a sweep cannot take them; and/unic-archon-dlc:specswrites a PRD file rather than a GitHub spec issue, so the Consumer work does not create the hazard either.orioltf commentedon Aug 14, 2026
Close-list, re-checked against
developat5366a4a— awaiting approval, nothing closed40 open non-wayfinder issues. Same shape as the draft: 33 close, 7 keep. Four verdicts keep their
outcome but change their reason, and one moves between groups. Each change is evidence the draft did
not have, because #374 merged in between.
Close — 33
slopcheckflag · #355validateConfig· #371artefact-install· #300 a new test over prose · #309 the provider-agnosticism barrierfirst installline · #302 scratch-Consumer clone —DXP-DesignSystemis the Consumer now5366a4a/cleanup· #282 wayfinder as a Boxunic-ticket-specification, shipped and running inDXP-ProfileServicesKeep — 7
unic-pr-review— a separate live product, in use at the Consumer today. Outside this map, in nobody's freezerresolvedorclosed. On Azure DevOps with a client watching the board, a stale work-item state is a product defect, not processBacklog goes 40 → 7.
What changed from the draft, and why
#362 — same verdict, different reason. The draft closed it as frozen process work. Two of its
three divergences no longer exist: the pre-audit section that read a stale tree is deleted, and the
needs-amendmentskip path whose end was undefined is deleted with it. The third survives, on Gate 3— an auditor whose verdict decides a merge, with nothing saying its load-bearing claims are re-run.
That is a design call with a real cost, which is why it closes rather than becoming a ticket.
#372 — same verdict, stronger reason. Not frozen: the 121-character marker was the audited-commit
comment, and #374 removed it. There is no longer a query looking for it. Closes as subject removed,
which is a different thing from not fixed.
#365 — the absorption is now a fact, not a forecast. #374 amended all seven sites. Of its ten
criteria, one was met by being deliberately reversed (the state-roles cell), and two are moot.
Both are recorded on the issue.
#303 — the reason improves, and this one is worth a look. The draft closed it as frozen process.
The hazard is also currently not live: this map's tickets carry
wayfinder:*and noready-for-agent, so a sweep cannot take them, and/unic-archon-dlc:specswrites a PRD file ratherthan a GitHub spec issue. #386 also stopped
triage-labels.md's table contradicting the paragraphbelow it. It stays a real hazard the day anyone runs
/to-specagainst this tracker again, andclosing it accepts that.
#309 moved groups. The draft absorbed it into #380. Its barrier is
test/box-staging-and-repo-pinning.test.mjs, which #381 deletes outright — so it dies with the tests,not with the quality-bar question. #380 is where its replacement gets decided.
Stream check
Each stream proposed for closing has no child that stays open:
/setupcorrectnessFour the maintainer decides explicitly
Unchanged from the draft. Each is real work that closes because it does not block the Consumer, not because it
is wrong:
ready-for-agentlets an unattended run re-dispatch it;a killed Archon run can commit work and open no PR. Both are genuine night-shift hazards, and night
shift is the speed mechanism. bug(repo): a merged issue keeps ready-for-agent, so an unattended run can re-dispatch it #345 has grown sharper since the draft, not blunter: with the
pre-dispatch audit gone, nothing at all inspects an issue before dispatch.
config-schema.mjsforces the defaults into comments anyway. Real work eitherway, folded into Delete lib/ and test/ — take the plugin to zero code #381.
because Rewrite commands/setup.md from scratch as prose #383 and Prune the ADRs aggressively #384 make it wrong again; the rewrite belongs after them.
real evidence at the Consumer.
On approval
Say the word and each closed issue gets a comment naming this map and the reopen path, and the map's
Out-of-scope section grows to list them. Approve the list as a whole, or name the ones you want moved.
orioltf commentedon Aug 14, 2026
Resolution — approved and executed
The maintainer approved the close-list on 2026-08-15. 33 issues closed, 7 kept. The non-wayfinder
backlog is 40 → 7, verified by re-querying after the pass.
Each closed issue carries a comment naming this map, its specific reason, and the reopen path: a scope
call against the destination, not a judgement that the issue is wrong; if it turns out to block the
The Consumer work it returns as a fresh ticket, not a resumption.
#256closed as completed, everything else as not planned. That distinction is load-bearing —unic-ticket-specificationshipped and runs inJura.ECommerceandsitecore-website, and recordingit as out-of-scope would lose that.
What still stands open
resolvedorclosed. On a client-visible Azure DevOps board, a stale work-item state is a product defectunic-pr-review— a separate live product, in use at the Consumer today, outside this mapOne deviation from the acceptance criteria, deliberate
The last criterion says every closed issue is listed in the map's Out of scope section. Twelve of
the 33 do not belong there, and putting them there would corrupt what that section means. The map
defines Out of scope as work ruled past the destination; these twelve sit on the route:
They are recorded on the map in their own paragraph, immediately under Out of scope, saying exactly
that. Twenty-one issues are listed as genuinely out of scope, grouped by reason.
If you would rather have all 33 in one list, say so and I will merge them — but the map would then
claim that finished work and absorbed work were ruled beyond the destination, which is not what
happened to them.
Two things the pass surfaced that are not decided here
inspects an issue before dispatch at all, so a merged issue keeping
ready-for-agentis now theonly thing standing between an unattended run and re-dispatching finished work. It closes because it
does not block the Consumer. That is the trade decision 20 already took, applied consistently.
wayfinder:*, and/unic-archon-dlc:specswrites a PRD file rather than a GitHub spec issue. Theday anyone runs
/to-specagainst this tracker again, it is live.Neither is fog: both are recorded consequences of decisions already taken, and neither needs a ticket
unless the Consumer work hits them.