Deferred, not rejected. Promote by moving an entry into BACKLOG.md unchanged; the BL-#### space is
shared, so promotion needs no renumbering.
methodology/The design, so it is not re-derived. Two ways to run a backlog, differing in exactly one dimension:
| Element | Cascade mode | Focused mode |
|---|---|---|
| Upstream linkage | strategy → pillar → epic → item | none required |
| Epic charter, exit criteria | required | skipped |
Item format, frozen intent, Status/Test, DoD gates, lock, Code Map at M+ |
required | identical |
Nothing that protects correctness is relaxed. What is dropped is a charter the work was never going to earn. Mode is a property of the item, not the project: a team runs the cascade for its roadmap and intake for interrupts, in the same backlog, in the same week.
One hard rule blocks the obvious version. “Items live in exactly one epic; new items always go into a
specific epic” is in 00’s hard-rules table, so allowing epic-less items is a hard-rule amendment and a
MAJOR bump. The cheap route — the one E00 is running now — is a standing intake epic:
the rule stays literally true, filing takes seconds, and nothing published changes. The expensive route is
to amend the rule, which is conceptually cleaner and costs a major version plus a changed rule adopters
have already pasted into their own instruction files.
The part worth publishing is not the mode — it is the measurement. Report the share of closed items that arrived through intake rather than the cascade, and read a high ratio as evidence against the cascade. If most real work bypasses four planning layers, those layers are not earning their cost and should be trimmed rather than enforced harder. This methodology has no other mechanism that can tell it a layer is not worth its cost — it took an external auditor to notice that nine of sixteen conventions had never been used.
Source: maintainer, 2026-08-25 — requirements churn and short-notice tickets make the four-layer cascade a tax on work that will not outlive the sprint.
Target: methodology/03_epics.md (the standing-epic shape) and methodology/04_backlog_items.md (the
mode as an item property), plus a line in 12 for the ratio report.
Why it is deferred rather than chartered: it would be the seventeenth convention added since v1.25.0, and only seven of the first sixteen have ever been exercised. This one gets used before it gets written — that ordering has never once happened in this project, and doing it in the other order is what produced the finding class the last release repaired.
Symptom that would justify promoting it: E00 has held real items through at least one semi-annual
pass, the intake ratio has been reported once, and the eviction rule has fired at least once — meaning
intake demonstrably feeds the cascade rather than replacing it. Absent that, publishing it is a guess with
a table around it.
Ranked below Agent File (BL-0058) and recorded so the next pass does not re-triage the same list.
BL-#### backlog, in code rather than markdown. Would inform F-01’s coordination question from the
side that actually built it. Promote if: the shared-ref lock protocol gets exercised and proves
awkward.05 explicitly declines. Worth citing by name in 05’s variants section
so the declined alternative has a concrete referent instead of a category.Rejected outright, with reasons, so they are not reconsidered: MemPalace and OpenViking (runtime memory
services against a markdown memory layer — the trade 08 already refuses); DSPy (would treat
ROLE_BRIEFS.md as optimizable artifacts; interesting, wrong axis, needs a runtime); and the entire
inference, training, serving and data-pipeline majority of the list, which governs a different layer of
the stack than this methodology does.
Method caveat, recorded because it bounds every verdict above: the triage ranked candidates from one-line README descriptions and opened no repositories.