Report Lifecycle — Consolidate, Then Archive

Every /wb* command writes one dated file per day per scope. After a month of work a scope holds thirty plan files, twenty audits and a dozen idea files — of which exactly one of each is live. This page is about the other twenty-nine.

---

The problem: a folder full of decoys

packages/wb-core/.wb/workflows/reports/2026/
├── 07/25/plans/plan_wb-core_20260725.md     ← 3 open tasks
├── 07/26/plans/plan_wb-core_20260726.md     ← 1 open task
├── 07/31/plans/plan_wb-core_20260731.md     ← all closed
├── 08/04/plans/plan_wb-core_20260804.md     ← 2 open tasks
├── 08/05/plans/plan_wb-core_20260805.md     ← all closed
├── 08/07/plans/plan_wb-core_20260807.md     ← 4 open tasks
├── 08/08/plans/plan_wb-core_20260808.md     ← 1 open task
└── 08/09/plans/plan_wb-core_20260809.md     ← today

Nothing here is wrong. Every one of those files was correct on the day it was written. But ask the practical question — "what is open in wb-core right now?" — and there is no file that answers it. The answer is spread across five of them, and the newest file is not a superset of the others.

Worse, both a human and an agent will reach for the wrong one. Open plan_wb-core_20260807.md because it is the one you remember, and you work from a backlog where four tasks are open that were closed two days ago. The file does not look stale — it looks like a plan.

A stale file is more dangerous than a missing one. A missing file makes you go looking. A stale one answers confidently and wrongly.

---

The two halves

The fix has two halves, and the relationship between them is the whole point of this page.

Half	What it does	Reversible?	Trigger
Consolidate	absorb every still-open item from every older file of this category into today's file, then repair that file's structure and links	n/a — purely additive	passing an existing output file with no other flags
Archive	move the now-superseded <DD>/<category>/ folders out of reports/ into archives/	yes, per folder	--archive, never implied

/wbPlan  <plan_file.md>              # half one: consolidate
/wbPlan  <plan_file.md> --archive    # half one, then half two

Consolidate: make one file worth reading

Passing an existing output file with no other flags means "make this the one file I have to read."

1. Repair in place — the full self-correct pass: link beautification, canonical hrefs, missing sections inserted in canonical order, Requires column and Action Types legend present, derived blocks re-synced and the sync oracle run.
2. Absorb — walk the scope's entire reports/ tree for prior files of the same category and merge every still-open item into this one.
3. Recompute — the 🌊 matrix, the how-to-run block, and What's Next, over the merged set.

What counts as "still open" is per-category, because the categories mean different things:

Category	An item is still open when…
plans	☐ Done or ☐ Valid is ⬜ or 🔨 — never ⏸️ Deferred / 🚫 Cancelled, those are decided
audits · reviews · security	the finding is 🔴/🟡 and no later file records it resolved
ideas	status is not 🎯 Promoted, 🚫 Rejected or ✅ Shipped
visions	the proposal was never promoted to an idea or a plan task
tests	the case is failing or skipped

Two rules do the real work here:

• Merge, don't concatenate. A task carried forward for four days appears once, bearing its oldest origin link — not four times.
• De-duplicate on the item's text, not its ID. IDs are per-file and restart at 1, so de-duplicating by ID silently collapses four unrelated tasks into one. This is the single most destructive way to get consolidation wrong.

Absorbed rows keep their wording, Requires, Dep, priority and estimate, and gain an origin link:

| 7 | 🔨 Worker | 4 | Port the WBChart renderer *(from `plan_wb-core_20260805.md`)* | … |

The source files are never modified. Consolidation copies. Only --archive moves.

Archive: retire what is now redundant

Once consolidation has run, the older folders hold nothing that is not also in today's file. --archive moves them:

<scope>/.wb/workflows/
├── reports/<YYYY>/<MM>/<DD>/<category>/     ← live: the newest date per category
└── archives/<YYYY>/<MM>/<DD>/<category>/    ← history: same path, same depth

---

🔴 The ordering invariant

Consolidate always runs before archive, in the same invocation, and archive is skipped if consolidation did not complete.

Archiving first does not delete an open task — it makes it invisible. The task still exists, on disk, in a file nobody will open: nothing live references it, and the next /wbStandup no longer scans the tree it sits in. Six weeks later you rediscover it and cannot tell whether it was dropped on purpose.

This is the one destructive failure mode of the whole feature, and it is why:

• archiving is opt-in (--archive) rather than automatic;
• no other flag implies it;
• a self-correct pass never archives — repair is safe to run unattended and runs often; moving folders is not and should not;
• the CLI's dry-run prints a (kept: …) column, and a row reading ⚠️ no current file in this category is the signal that consolidation has not run. Archiving anyway is exactly the failure above.

---

Three properties of the archive tree

Each of these is load-bearing. None is aesthetic.

1. Same depth as reports/

archives/ is a sibling of reports/ under .wb/workflows/, and the <YYYY>/<MM>/<DD>/<category>/ path below it is identical to the one the folder had while live.

.wb/workflows/reports/2026/08/07/plans/plan_x_20260807.md    ← 7 dirs to scope root
.wb/workflows/archives/2026/08/07/plans/plan_x_20260807.md   ← 7 dirs to scope root

Every relative href inside a moved file that pointed at something moving with it still resolves after the move, unchanged and unrewritten. A plan's task 1 link works from the archive exactly as it worked live. A shorter or prettier archive location would break every link in every file it stores.

2. The unit is the folder, never the file

A plan's tasks/, waves/ and explanations/ live inside <DD>/plans/, beside the .md:

2026/08/07/plans/
├── plan_x_20260807.md
├── tasks/task_1/task_1_report_x_20260807.md   ← the plan links here
├── tasks/waves.md
└── waves/wave_A_x.sh

Move the .md alone and every task report it links to is orphaned. So <DD>/<category>/ moves whole, or not at all.

3. The original date is preserved

A plan dated 20260807 archives to archives/2026/08/07/, not to today's date. The archive records when the work happened, not when someone got round to tidying.

---

What a move leaves behind

Each archived .md is stamped once, immediately below its front-matter (never above it — a banner before --- breaks the YAML parse):

> 🗄️ **ARCHIVED 2026-08-09** — this file is history, not live state.
> Superseded by `plan_wb-core_20260809.md`.
> Moved from `.wb/workflows/reports/2026/08/07/plans` by `wb-flow archive`. …

and one row is appended to archives/archive_log.md:

Date	Category	Archived from	Archived to	Superseded by
2026-08-09	plans	…/reports/2026/08/07/plans	…/archives/2026/08/07/plans	plan_wb-core_20260809.md

So a reader who lands on an old file via a stale bookmark knows immediately that it is history and can reach its replacement in one click — and every move is individually reversible:

wb-flow archive --restore=.wb/workflows/archives/2026/08/07/plans

---

The two exemptions: /wbStandup and /wbTrack

Every other daily-history command keeps one live file and retires the rest. standups/ and tracks/ are never swept, and this is not an oversight waiting to be fixed.

A standup's entire content is "here is what was open yesterday, here is today." A track is the session narrative. Both derive their value from the series — the eighth consecutive standup listing the same blocker is the signal; that is the whole reason to read it. Keep only the newest and you have thrown away the thing that made it worth reading, while keeping the one entry that says least.

So the series stays whole, and /wbStandup takes the inverse role instead.

/wbStandup --archive — the fleet-wide sweep

/wbStandup <monorepo-root>/            # unchanged — the daily agenda
/wbStandup <monorepo-root>/ --archive  # …then one live file per category, per scope

It is the right command to own the sweep because it has already done the hard part. PHASE 1–3 build the complete inventory of what is open across every scope — and that inventory is what makes the sweep safe, because it is the cross-check:

Every open item listed in the agenda must now appear in exactly one live file. An agenda item with no live home means consolidation missed it — stop before sweeping.

Because it spans every scope rather than one, it always previews and asks before applying. A per-command --archive moves a handful of folders; this one can move a hundred, and per-folder reversibility is little comfort at that volume.

---

The CLI

Templates shell out to wb-flow archive rather than hand-rolling mv — the same rule that makes wb-flow snap the only writer of pin symlinks, except this one moves work product, so the stakes are higher.

wb-flow archive <scope> --type=plans --dry-run   # always preview first
wb-flow archive <scope> --type=plans             # apply
wb-flow archive <plan_file.md>                   # a FILE pins itself as the keeper
wb-flow archive <scope> --recursive --all        # every scope below, every category
wb-flow archive --list                           # what is already archived
wb-flow archive --restore=<folder>               # undo one move

Flag	Effect
--type=<a,b>	Categories to sweep (default: every non-exempt one found)
--keep=<N>	Keep the N newest dates per category (default 1)
--before=<YYYYMMDD>	Only archive dates strictly before this one
--recursive / -R	Sweep every scope with a .wb/ below the target (fleet mode)
--include-standups · --include-tracks · --all	Opt the exempt categories in explicitly
--no-banner · --no-log	Skip the stamp / the log row
--restore=<path> · --list	Undo one move · show what is archived
--dry-run / -n · --json	Preview · machine-readable

It resolves the .wb/ root, refuses to touch the newest folder in each category, refuses to clobber an existing archive entry, skips the exempt categories, prunes the empty <YYYY>/<MM>/<DD> shells it leaves behind, writes the banner and the log, and exits non-zero on a bad target. An agent reimplementing this with mv gets none of that, and its first mistake costs a plan.

Passing a file pins the keeper

wb-flow archive .wb/workflows/reports/2026/08/05/plans/plan_x_20260805.md

The named file becomes the keeper for plans/: everything older is archived, everything newer is left alone. Sweeping forward from a hand-picked keeper would destroy work the caller never mentioned — so it doesn't.

---

Reading a dry run

📋 DRY RUN — nothing will move

  packages/wb-core/
   audits      .wb/workflows/reports/2026/08/04/audits
               → .wb/workflows/archives/2026/08/04/audits   (kept: audit_wb-core_20260807.md)
   plans       .wb/workflows/reports/2026/08/07/plans
               → .wb/workflows/archives/2026/08/07/plans   (kept: plan_wb-core_20260809.md)
   · standups: exempt — it is the log (use --include-standups or --all)

   2 folder(s) would move.

Two things to check before you apply:

1. The (kept: …) name is the file you expect to be live. Note that audits and plans resolve different keeper dates — keepers are per-category, so a category with no recent activity keeps its own newest file rather than being dragged out by a busier neighbour.
2. No row reads ⚠️ no current file in this category. That means nothing live supersedes the folder, which almost always means consolidation has not run.

---

Related

• self_correct_mode.md — the repair half, and why it never archives
• plan_state_management.md — which states count as "open"
• cli-subcommands.md — wb-flow archive beside the other subcommands
• live_reports_snapshot.md — how the reports tree is laid out

---

← Concepts · Home | flow.wbc-ui.com