feature-ideation enhancement backfill + dry-run + marker continuity (prereq to deprecating idea-enhancer)

[don-petry]

Summary

The idea-enhancer → feature-ideation consolidation (#872) folded the new-idea
enhancement path but dropped two idea-enhancer capabilities — the existing-backlog
sweep and the dry-run — and the idempotency marker differs. Restore them as
first-class feature-ideation modes before idea-enhancer is deprecated (#876), so the
existing Discussions that never got ideated can be enhanced safely and repeatably.

Background — verified gaps (2026-06-25)

• No backfill sweep. feature-ideation-reusable.yml enhances only a single idea via
  target_discussion (fired by discussion:created). There is no mode that sweeps the repo's
  open, human-authored, not-yet-enhanced Ideas backlog — the capability idea-enhancer
  provided via scripts/idea-enhancer/{gather-candidates,post-enhancement}.sh + a weekly sweep.
• No enhancement dry-run. feature-ideation's dry_run input covers generation, not
  enhancement-of-existing, so a backfill can't be previewed.
• Marker mismatch. feature-ideation skips on <!-- feature-ideation:enhanced -->
  (reusable lines 308-310); idea-enhancer wrote <!-- idea-enhancer:enhanced -->. After
  cutover, every idea idea-enhancer already enhanced would be re-enhanced — the exact
  failure [Phase 2] Validate the folded feature-ideation enhancement mode on real .github-private Discussions (next ring) #875's Dev Notes warn against.

Goals

1. Marker continuity (E1). feature-ideation's enhancement idempotency check recognizes
   both <!-- idea-enhancer:enhanced --> and <!-- feature-ideation:enhanced --> as
   already-enhanced; new comments emit one canonical marker. No re-enhancement across cutover.
2. Backfill sweep mode (E2). A workflow_dispatch input (e.g. enhance_backlog: true)
   makes feature-ideation sweep the target repo's open, human-authored, not-yet-enhanced
   Ideas Discussions and post exactly one structured enhancement comment each
   (sharpened problem / context / impact-effort / acceptance criteria), idempotent via the
   marker; bot-authored and already-enhanced ideas are skipped. Ports idea-enhancer's sweep.
3. Dry-run the backfill (E3). dry_run=true + backfill logs the intended per-Discussion
   enhancement comments to the JSONL artifact and posts nothing — a safe preview of a backfill.

Acceptance criteria

• E1: a Discussion carrying <!-- idea-enhancer:enhanced --> is not re-enhanced by
  feature-ideation; bats/test coverage for both markers.
• E2: enhance_backlog dispatch posts one structured comment per open, human-authored,
  un-enhanced Idea (idempotent; bots + already-enhanced skipped); a second run is a no-op.
• E3: dry_run=true + enhance_backlog writes the intended comments to the artifact,
  posts nothing; the artifact lists one entry per candidate Discussion.
• Docs: ci-standards §10 / pipeline doc describe the backfill + dry-run modes and the
  operator runbook for backfilling an existing repo's Ideas backlog.
• Re-sequence: [Phase 2] Validate the folded feature-ideation enhancement mode on real .github-private Discussions (next ring) #875 validates the backfill (dry-run over the real .github-private
  backlog, then live); [Phase 3] Deprecate idea-enhancer in .github-private: remove workflow + scripts + test, update the pipeline doc #876 (deprecate idea-enhancer) is blocked until E1-E3 land.

Out of scope

• The weekly generation pipeline's prompt/model.
• Cross-repo ring promotion of feature-ideation (separate; feature-ideation/next already cut).

Companion to #872 (consolidation). Add idea:approved to hand to the BMAD Scrum Master.

[github-actions[bot]]

📋 Initiative planned by the BMAD Scrum Master (Bob).

Epic #934 — feature-ideation enhancement backfill + dry-run + marker continuity (restore idea-enhancer parity before deprecation)

3 stories created (inert — labelled initiative, NOT initiative:auto):

• [Phase 1] Expose the feature-ideation backfill input: sync the stub to forward enhance_backlog #935 (S) — [Phase 1] Expose the feature-ideation backfill input: sync the stub to forward enhance_backlog
• [Phase 1] Document the feature-ideation backfill + dry-run modes and the backfill operator runbook #936 (M) — [Phase 1] Document the feature-ideation backfill + dry-run modes and the backfill operator runbook
• [Phase 2] Validate the feature-ideation backfill on the real .github-private Ideas backlog (dry-run then live) #937 (M) — [Phase 2] Validate the feature-ideation backfill on the real .github-private Ideas backlog (dry-run then live)

Open questions for review:

• Re-sequence [Phase 3] Deprecate idea-enhancer in .github-private: remove workflow + scripts + test, update the pipeline doc #876 (deprecate idea-enhancer): it must NOT start until E1-E3 land and the backfill is validated (Story 3). This plan cannot add a blocked_by edge onto the pre-existing [Phase 3] Deprecate idea-enhancer in .github-private: remove workflow + scripts + test, update the pipeline doc #876 — a maintainer should manually set [Phase 3] Deprecate idea-enhancer in .github-private: remove workflow + scripts + test, update the pipeline doc #876 blocked_by this epic's backfill-validation story (and keep [Phase 3] Deprecate idea-enhancer in .github-private: remove workflow + scripts + test, update the pipeline doc #876 at initiative:hold) until the go/no-go note confirms marker continuity + single-enhancement.
• Should these stories attach to existing consolidation epic Initiative: Consolidate Discussion-triggered ideation into feature-ideation, deprecate idea-enhancer, and ring-release fleet-wide #872 instead of a new epic? feature-ideation enhancement backfill + dry-run + marker continuity (prereq to deprecating idea-enhancer) #933 is framed as a companion restoring capabilities Initiative: Consolidate Discussion-triggered ideation into feature-ideation, deprecate idea-enhancer, and ring-release fleet-wide #872 deferred; a separate epic keeps the human go/no-go gates independent, but a maintainer may prefer to fold them into Initiative: Consolidate Discussion-triggered ideation into feature-ideation, deprecate idea-enhancer, and ring-release fleet-wide #872's DAG.
• For the stub enhance_backlog input (Story 1): is the central standard stub template (petry-projects/.github standards/workflows/feature-ideation.yml) being updated in lockstep, or should the .github-private stub input be recorded as a documented repo-specific exception in AGENTS.md so a template sync does not reset it (as with the existing feature-ideation/next sanctioned-pin exception)?

---

Review the epic and its sub-issue DAG, adjust as needed, then add initiative:auto to epic #934 to hand it to initiative-driver for auto-implementation.