Skip to content

FIX-1663: amend closure QA plan for Shift Manager and FIX-1690 - #2592

Merged
jhoffner merged 4 commits into
mainfrom
spec/FIX-1663
Oct 1, 2026
Merged

jhoffner merged 4 commits into
mainfrom
spec/FIX-1663

Conversation

@jhoffner

@jhoffner jhoffner commented Oct 1, 2026 •

Copy link
Copy Markdown
Collaborator

The problem

The closure QA plan for the Workforce lab shell epic was merged (#2426) before the shell was renamed and before FIX-1690 shipped. Run as written, it fails on things that aren't findings:

  • paths that no longer exist (labs/app-lab, goals/app-lab/)
  • a step that types into a session the coder seat only has while a task is running
  • an Inbox reply that DevForce's EM can't take
  • a leg c sweep that names chrome Shift Manager doesn't mount
  • a pentest config it would write over a file already on main
  • a "no org is refused" step that the dev org now answers

The closure re-run reads this spec, so the spec has to match main first.

What this does

This is a post-merge amendment of specs/issues/FIX-1663/. Every change was already decided elsewhere. This PR records each one with its source, in EVOLUTION.md.

  1. Names. The shell is now Shift Manager (FIX-1649: rename app-lab to shift-manager, add day/night boot shift and DevTeam profile #2588):

    • it lives in labs/shift-manager, with its checks under goals/shift-manager/
    • the DevForce tree opens with --team devteam
    • light and dark come from --shift day|night

    The commands in the legs run on main at 0c49e70. The DevForce tree keeps goals/devforce-lab/lab/. Quotes in Notes from review keep the old name.

  2. Leg c sweep. Leg c now sweeps what Shift Manager promised to draw: the message, reasoning, tool, code block and ask cards, plus the devtool page (EM call with FIX-1688/1689, fix(app-lab): load the design-system theme, draw the Session with registry copies (FIX-1688, FIX-1689) #2532). The navigator, roster, board panels, seat detail, task plan card and the twelve FIX-1655 fixes are out.

  3. a4. The token goes into eng.coder's running task session, through FIX-1690's door (FIX-1690 (stop): a block can stop another request in its own session #2538, FIX-1690 (door): a person's message continues a coding run's session #2544, FIX-1690 (lab): App Lab sends a person's line into a worker's session #2565).

  4. Part 4. The Inbox reply is graded on FIX-1690's test seat. On DevForce's EM, the reply box is disabled and its line says why (FIX-1690's second fork, spec(FIX-1690): a person can't send a message into a worker's session from App Lab #2525).

  5. a2. Runs with no --devtool. Shift Manager serves its own devtool in the same process (fix: Open trace opens the task's run, and the DevForce Lab's approved run completes (FIX-1691, FIX-1692) #2531).

  6. b0 compares instead of writing. goals/pentest-lab/lab/fsdev.config.mts has been on main since FIX-1693 · FIX-1694 · FIX-1696: App Lab says what a Lab's config provides, refuses a Lab with no organization, and reaches a DevForce document #2529. The writer still works from Shift Manager's README alone, but it now writes its copy to a scratch location, and the run compares that copy with the committed file. A semantic difference is a finding. The closure PR no longer carries the config.

  7. Leg b checks that a Lab always opens under an org. The epic promises no org-less mode (DESIGN.md, Org switcher; the epic SPEC's Not done if). The dev org __fsd_default_org__ counts as an org: FIX-1442 sanctioned it, and the EM call on FIX-1693 · FIX-1694 · FIX-1696: App Lab says what a Lab's config provides, refuses a Lab with no organization, and reaches a DevForce document #2529 kept it. So leg b now checks three things: a Lab with no resolver opens under the dev org, the org switcher names it, and no screen renders without an org or as an empty app. A new scratch control, no-org, renders the switcher with no org and must make that check fail.

The goal, D1 and the rules for findings are unchanged. The biggest change is that this narrows the epic's pinned leg c sweep for the closure without editing the epic itself; EVOLUTION records it, and the epic text is reconciled at epic wrap.

What's asked of you

Nothing beyond review. Every change above was decided before this PR. Items 6 and 7 are engineering calls by the epic coordinator, and the anti-game exception for teams/devteam/fsdev.config.mts (the DevTeam profile #2588 added) follows from adopting --team devteam.

Parts worth reviewing closely

  • PLAN.md → Checks (a2, a4, b0, b, c), the leg c sweep paragraph, Controls (no-org)
  • PLAN.md → Part 4, the ER-15 row
  • DECISIONS.md → D2 Locks in
  • EVOLUTION.md (every amendment and its source)

Spec: specs/issues/FIX-1663/ · Linear FIX-1663 · epic FIX-1649 · original review #2426 · amends after FIX-1690 (#2525), #2529 and #2588 · also cross-closes FIX-1690's EVOLUTION row for this amendment · merges after direction approval and required checks.

How to review this: altitude, what's in scope, what's deliberately unsettled

This is a direction set, not an implementation. Reviewing it well means
answering one question: is this the right approach?

In scope to challenge:

  • The goal and the problem framing in SPEC.md — is the goal the real need at full size,
    for the people named, or a smaller goal that would let us call this done early? Would the
    named check fail if the goal were not met?
  • The approach — will it work, does it fit the architecture and docs/philosophy.md?
  • Any numbered decision in DECISIONS.md — that's the sign-off surface.
  • A case BUSINESS-RULES.md misses, or a rule that would invalidate the design.
  • Scope — a deliverable that shouldn't ship, or one that's missing.
  • The reader-facing promises in DOCS.md and their fit with the rules; a missing or
    misleading lineage claim in conditional EVOLUTION.md.

Out of scope — deliberately unsettled here, and owned by the implementing agent:

  • Names, signatures, file paths, and module layout beyond the few PLAN.md pins on purpose.
  • Local structure: which helper, how a function is decomposed, error-message wording.
  • Micro-optimizations and style preferences.
  • Test names and internal test structure (the behaviours are the rules; the tests are not).
  • Anything PLAN.md left open on purpose — it is directional by design, and a gap at that
    altitude is intended, not an omission.
  • The figures, at the pixel level. A figure is right when it reads right. A misplaced label
    or a colour you'd have chosen differently is not a finding; a figure that shows a shape the
    decisions don't say is.
  • The solution sketch, at the line level. PLAN.md may include a rough pseudocode sketch
    showing the shape of the proposed solution. It is knowingly incomplete, is not real code,
    and is not what will ship. Review it for directional viability only. Do not report
    that it lacks error handling, omits edge cases, names things that don't exist in the repo, or
    wouldn't build: all of that is true on purpose.
  • POC implementation polish. Experiments under the owning spec's poc/ are retained,
    not production code. Read them for directional evidence, not production completeness.
    Their isolation from production/default discovery and absence of secrets or generated
    dependencies are in scope; retention is not permission to weaken repository checks.

One request, if you are an automated reviewer: the two lists above are the whole difference
between a useful review of this document and a long one. Volume is not signal here — a single
finding above the line is worth more than twenty below it.

Feedback in the second list is welcome and gets recorded verbatim in PLAN.md → Notes from review for the implementer to weigh against real code. It will not be argued with, and it
will not be folded into the design prose. Please don't re-raise it: one mention is enough.

The one exception is the sketch and the POC. Line-level feedback on either is dropped, not
recorded — there is nothing for an implementer to weigh about code that isn't shipping. Feedback
about their direction is not line-level and is treated like any other above-the-bar finding.

Direction approval does not waive merge-time required checks, approvals or thread policy.
Optional comments do not require another design round merely to reach zero comments.

Verification: paths, commands, validator
  • Every repo path cited in the folder was checked with ls on this branch, which is based on origin/main 0c49e70. Three don't exist, on purpose. goals/shift-manager/a-lab-is-worked-through-one-skinned-shell/ is S1, which the closure run writes. labs/app-lab and goals/app-lab/ appear only in EVOLUTION and the review quotes, as history. The no-resolver Lab, labs/shift-manager/test/fixtures/ask-lab/fsdev.config.mts, falls back to DEFAULT_ORG_ID = __fsd_default_org__ (packages/core/src/types/auth.ts) when it is booted with no bearer. Commit 46bb201, which added the pentest config, is on main.
  • Flags checked against labs/shift-manager/bin/start.mts: --team, --config, --devtool, --shift.
  • Env var and controls checked: DEVFORCE_LAB_HARNESS=claude-code is in goals/devforce-lab/lab/harness.mts, and GOAL_CONTROL=static-names|optimistic-post are in goals/shift-manager/it-opens-a-lab/controls/.
  • pnpm build:assets is in the root package.json.
  • node scripts/validate-spec-folder.mjs → ✓ legacy spec folders are clear and no dangling legacy spec paths are cited (exit 0), after every push.
  • figures/d2-pentest-config.svg was rendered in headless Chromium after the text change, and the label fits its box.

🤖 Generated with Claude Code

https://claude.ai/code/session_01Cnm5DbhFmkarpesBpPQ9Dq

Post-merge amendment of the retained closure spec (original review #2426):
- names, paths and commands follow the Shift Manager rename (#2588)
- leg c sweeps the promised set: message, reasoning, tool, code block and
  ask cards, plus the devtool page (#2532)
- a4 targets eng.coder's running task session (FIX-1690: #2538/#2544/#2565)
- Part 4's Inbox reply is graded on FIX-1690's test seat (#2525)
- a2 runs without --devtool; Shift Manager serves its own (#2531)

Recorded in DECISIONS -> How it got here, and a new EVOLUTION.md.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Cnm5DbhFmkarpesBpPQ9Dq
@chatgpt-codex-connector

Copy link
Copy Markdown

You have reached your Codex usage limits for code reviews. You can see your limits in the Codex usage dashboard.
To continue using code reviews, add credits to your account and enable them for code reviews in your settings.

@vercel

vercel Bot commented Oct 1, 2026 •

Copy link
Copy Markdown

The latest updates on your projects. Learn more about Vercel for GitHub.

3 Skipped Deployments
Project Deployment Actions Updated
flow-state-dev-docs Ignored Ignored Oct 1, 2026 5:53pm UTC
flow-state-dev-kitchen-sink Ignored Ignored Oct 1, 2026 5:53pm UTC
flow-state-dev-ui Ignored Ignored Preview Oct 1, 2026 5:53pm UTC

Request Review

@linear-code

linear-code Bot commented Oct 1, 2026 •

Copy link
Copy Markdown

FIX-1690

FIX-1663

jhoffner commented Oct 1, 2026

Copy link
Copy Markdown
Collaborator Author

Second look

Verdict: Appropriately scoped. The amendment is mostly a rename plus two real narrowings (leg c sweep, a4 target). The extra surface is its own paperwork. Spec-only PR (8 files under specs/issues/FIX-1663/), so there is no code to refactor and nothing for an architecture pass to deepen. Estimated line deltas below are for the spec text.

# Finding Recommendation Est. Δlines Sev Conf
1 Part 4 "Token names" row, and leg b's "no file there names a pentest seat" clause, re-grade what labs/shift-manager/test/static.test.ts already enforces in CI (no literal colour, no seat/team/channel names; its tests plant a violation to prove they can fail). Drop the row and the clause. Cite P3.3's pnpm test in the report instead. A run-time grep would be a second copy that can drift. −20 / +2 Low Med
2 The same amendment is recorded three times: DECISIONS "Amendment after merge", the EVOLUTION table, PLAN's "As amended after merge" paragraph. Calls 1–3 appear nearly verbatim as EVOLUTION rows 3, 1, 2. Keep EVOLUTION (the template requires it). Cut DECISIONS items 1–3 to one-line pointers, keep 4–5 (EVOLUTION doesn't carry them), and trim the PLAN paragraph to a link. −15 / +3 Low High
3 Coherence: PLAN's header and "At implement time" still say "FIX-1662's are in review as #2424", "FIX-1664 has no spec yet", and cite a no hit on main at 904c81b78 grep. These contradict the amended preamble (commands now run on main at 0c49e70). Reconcile those lines (~6) so the closure worker doesn't read two states of the world. This is the one finding worth doing before merge. ~6 Med High

Considered and deemed justified

  • Narrowed leg c sweep. It removes checks for chrome Shift Manager never mounts. It is the one change that removes scope.
  • hardcoded-accent control, although FIX-1655 has a same-named check. It fails a different leg (no-theme leg c) and proves that leg can go red, which BP-003 requires.
  • a4 retargeted to eng.coder's running task session, plus QR-9. This applies FIX-1690's merged door and adds no surface.
  • Rename edits across BUSINESS-RULES, DECISIONS, DOCS, SPEC and the two SVGs. Mechanical, and BP-034 requires them.
  • D2's README-only isolated writer. Unchanged, and it is the "builds the next Lab" journey.

On the two open conflicts in the PR body

I agree with the author's recommendation on both and don't see a better shape.

On the anti-game exception for teams/devteam/fsdev.config.mts: it is the narrowest fix that lets --team devteam stand, and it's named explicitly. No objection.

No architecture-refactor opportunities. A spec amendment is not a place to restructure code.


Generated by Claude Code

- b0's README-only writer writes the pentest config to scratch and the run
  compares it with the file #2529 committed; a semantic difference is a
  finding. The closure PR no longer carries S2.
- Leg b: a Lab with no resolver opens under the dev org (FIX-1442, EM call
  on #2529), which the org switcher names; never an org-less or empty app.
  New scratch control `no-org` (switcher empty) must fail it.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Cnm5DbhFmkarpesBpPQ9Dq

@cursor cursor Bot left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Problem

The closure spec on main still names App Lab, the old leg‑c sweep, and pre‑FIX‑1690 session semantics, while Shift Manager, FIX‑1690, and closure run #1 have already moved on. This PR is the right kind of fix: a post‑merge amendment with an EVOLUTION.md lineage file and updated PLAN checks.

Verdict (simplify)

Direction: keep it. Renaming to Shift Manager, narrowing leg c to cards Shift Manager actually draws, aligning a4/ER‑15 with FIX‑1690, and using --shift day|night instead of inventing a no‑theme product switch all reduce real closure cost and match shipped reality.

Where the approach went further than necessary: the amendment is narrated in four places (DECISIONS §“Amendment after merge”, EVOLUTION, the PLAN banner, SPEC header) instead of following the FIX‑1500 / spec‑template split: EVOLUTION = lineage, PLAN = executable checks, How it got here = pre‑merge review history only. That duplication is the main complexity smell—not the underlying decisions.

Cleaner shape (optional before or right after merge)

  1. Collapse the five numbered items in DECISIONS to a single pointer: “Post‑merge amendment: Evolution.”
  2. Drop the PLAN “As amended after merge” block; paths in tables are already current.
  3. Add EVOLUTION rows (or drop DECISIONS bullets) for a2 / no --devtool and Part 4 Inbox grading if they are meant to be cross‑spec lineage—not only plan detail.
  4. Fix the stale FIX-1664 has no spec yet line (it contradicts the rest of this PR).
  5. Shorten P3.2 to “each child’s merged goal check under goals/shift-manager/” with links to child PLAN.md#pinned-names (FIX‑1690 is the net‑new stack member worth naming explicitly).
  6. Part 4 ER‑15: one sentence + link to FIX‑1690 open-inbox; note for the implementer that P3.2 already runs FIX‑1690’s goal check so the live run should not duplicate that harness work.
  7. Follow‑up (not blocking): close the loop on FIX‑1690 EVOLUTION row 10 (“pending FIX‑1663 amendment”) and reconcile epic FIX‑1649’s pinned leg‑c sweep at epic wrap, as this PR already says.

None of the above changes what closure must prove; they make one authoritative read path for the closure worker.

Approval

Not approving in this pass: the stale FIX‑1664 line and four‑way amendment duplication are fixable in minutes and will otherwise confuse the first implementer read. Happy to approve after that trim or if you prefer merging as‑is and folding the dedupe into a tiny follow‑up amend.

Open in Web View Automation 

Sent by Cursor Automation: Review For Best Approach

Comment thread specs/issues/FIX-1663/PLAN.md Resolved
Comment thread specs/issues/FIX-1663/DECISIONS.md Resolved
Comment thread specs/issues/FIX-1663/EVOLUTION.md Resolved
Comment thread specs/issues/FIX-1663/PLAN.md Resolved
Comment thread specs/issues/FIX-1663/PLAN.md Resolved
Comment thread specs/issues/FIX-1663/PLAN.md Resolved
Comment thread specs/issues/FIX-1663/PLAN.md Resolved
Comment thread specs/issues/FIX-1663/SPEC.md Resolved
…59282)

- Lineage for all seven amendment items lives in EVOLUTION.md; DECISIONS
  "How it got here" ends with one pointer; PLAN banner dropped; SPEC header
  points to Evolution only.
- PLAN: stale FIX-1664 line fixed; P3.2 re-runs each child's merged goal
  check by its Pinned names (FIX-1690 new to the stack); ER-15 row points at
  FIX-1690 open-inbox and reuses P3.2; leg c rationale links Evolution.
- FIX-1690 EVOLUTION: cross-close the FIX-1663 amendment row.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Cnm5DbhFmkarpesBpPQ9Dq

@jhoffner jhoffner left a comment

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

FSD Architect — coherence only. Not owner approval / not the merge gate (shared jhoffner login looks like Jake). Direction stamp still needs Jake.

No silent two-direction fork. Post-merge amendment of specs/issues/FIX-1663/ only. Rename / a4 / Part 4 / a2 / b0-compare / no-org reading match already-merged sources (#2588, FIX-1690, #2529, FIX-1688/1689). EVOLUTION correctly records narrowing the epic's pinned leg c sweep without editing specs/epics/FIX-1649/ — that is a documented deferral, not a quiet product fork. FIX-1689 already names the missing react chrome. Epic wrap or a thin epic amendment still owes the reconcile so main's epic text and the closure plan don't disagree forever.

Agree with Second look #3: PLAN's opening and "At implement time" still say FIX-1662 is in review as #2424, FIX-1664 has no spec, and cite 904c81b78. That contradicts the amended 0c49e70 preamble — fix before merge so the closure worker doesn't read two worlds.

Watchlist, not a blocker here: draft #2584 (FIX-1706) is a second rename stream (goals/devteam-lab, dirty vs main). This PR pins goals/devforce-lab/ after #2588. Close or rebase #2584 before it re-stales these paths.

No Architect ask beyond that. No Jake ping — the epic deferral is already in the PR body.

The D1 gap check cited 904c81b and expected no ask and no board; both
were closed by FIX-1666 and FIX-1667 on main. Point at their merged
checks instead.

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Cnm5DbhFmkarpesBpPQ9Dq
@jhoffner
jhoffner merged commit 85caed1 into main Oct 1, 2026
13 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants