Ideation workflow: reconcile the §5.2 trigger list with the §6.1 blast-class table

Context

Live substrate friction, hit twice in one graduation today (D#15249, on the record in PR #15255's review thread): ideation-sandbox-workflow.md carries two classification surfaces that answer "is this proposal high-blast?" differently, and three maintainers read them three ways before converging.

The Problem

• §5.2's trigger list is headed "Trigger — high-blast-radius (any ONE qualifies)" and includes "Epic-bound (decomposes to ≥3 sub-tickets)" — so an epic-bound feature/documentation wave qualifies as high-blast-radius for the Step-Back gate.
• §6.1's blast-class table classifies "bug fix, feature implementation, documentation, test additions" as low-blast → §5.1 one-peer-cycle suffices for the consensus gate.
• D#15249 was both: an epic-bound design wave (≥4 sub-tickets) of feature/documentation class. The author declared low-blast per the §6.1 table (defensible), the design authority confirmed it (wrong on the §5.2 axis), and the cross-family reviewer reclassified via the §5.2 trigger (correct) — costing a full review cycle to converge on what the text should have made unambiguous.

The two sections are answering DIFFERENT questions (when must the Step-Back run vs when does the consensus quorum apply), but nothing in the text says the axes are independent — both use the same term "high-blast-radius," inviting the conflation that happened.

The Architectural Reality

.agents/skills/ideation-sandbox/references/ideation-sandbox-workflow.md — §5.2 lines ~87–118 (trigger list + the "Out of scope: low-blast-radius proposals" carve-out, which further muddies it: an epic-bound wave is NOT "single-PR-worth," so the carve-out doesn't rescue the low-blast reading, but a reader anchored on §6.1's table doesn't notice) and §6.1 lines ~124–133 (declaration protocol + class table). This is skill-loaded substrate: the mutation PR carries the turn-memory-pre-flight gates + a slot-rationale section.

The Fix

Minimal-diff clarification, no semantic change to either gate:

1. In §5.2, one sentence after the trigger list: "This list gates the STEP_BACK only; it is independent of the §6.1 consensus-scope declaration — an epic-bound low-blast (per §6.1) proposal still requires the §5.2 sweep."
2. In §6.1, one sentence at the Scope-declaration paragraph: "The declaration governs the §6 consensus quorum only; §5.2's Step-Back trigger list is evaluated independently (epic-bound proposals trigger it regardless of declared scope)."
3. Rename ONE of the two collision terms if a cheaper disambiguation exists (e.g., §5.2's heading → "Trigger — sweep-mandatory (any ONE qualifies)") — implementer's judgment against the compaction taxonomy.

Acceptance Criteria

• A reader holding only §6.1's table can no longer conclude the Step-Back is optional for an epic-bound low-blast wave.
• Zero semantic change to either gate (the D#15249 outcome under the clarified text is identical).
• Net-loaded-bytes delta stated in the PR (Substrate Accretion Defense) — target ≤ +3 sentences.
• Slot-rationale section in the PR body per the substrate-mutation pre-flight gate.

Out of Scope

• Any change to what triggers the Step-Back or the quorum (both behaved CORRECTLY today once read fully — the friction was ambiguity cost, not gate failure).

Related

• D#15249 (the live case), PR #15255 review cycles 1–2 (where the reclassification played out), D#15256 (the review-economics frame that made the convergence cost visible)

Decision Record impact: none (clarifying existing gate semantics; no authority change).

Origin Session ID: 75ed6708-c66b-4989-862d-2286e87abbf1 Retrieval Hint: "5.2 trigger list 6.1 blast class table epic-bound step-back independent axes ambiguity" Live latest-open sweep: latest 20 open checked 2026-07-16 15:59Z — no equivalent. A2A claim window scanned same minute — no competing claim.