Plan integration follow-on (D-E): age-aware nudge and a learner-issued retire/snooze for an overdue item #26

Origin

• Owner decision D-E (docs/architecture/plan-integration/HANDOFF-2026-09-16.md §2): "an overdue item unrelated to the plan must not sit as an alternate indefinitely. Fact: due score already grows +1/day (cap +30), so it overtakes the +12 bias in ~2 weeks; missing are an age-aware nudge line and a retire/snooze action for a due card (only backlog topics can be resolved today)."
• Evidence: rubric row 2 (receipts/now-rubric-2026-09-16.md), owner verdict yes — clear the overdue review first — with the note that an unrelated overdue item must be proposed explicitly or be retirable by the learner.
• Full proposal, written from the code on main a5b9f903: proposals/2026-09-16-overdue-nudge-and-retire.md. Follows on from Integrate Study Plans across Now, Web architect, and MCP #7 (plan integration); not a blocker of Integrate Study Plans across Now, Web architect, and MCP #7.

What exists today (read from the code)

• A due item is a study_progress row whose last_seen reaches an interval in REVIEW_INTERVALS (history/progress.py: 1, 3, 7, 14, 30 days); a struggling row is always due. There is no next_review column and no "not due" state — a row stays due, and stays a now candidate, until it is studied again (record_progress) or deleted.
• learning/decision.py scores a due candidate 100 + min(days_ago, 30) (+35 struggling, +15 learning); a plan-related candidate adds PLAN_RELATED_BIAS = 12. An unrelated item overtakes a plan-related one on age alone after ~2 weeks — silently: nothing tells the learner why it moved up, and nothing offers "I have moved on from this".
• resolved exists only for parked backlog topics (record_topic_progress(confidence="resolved") → parking.resolve_parked_topic), not for study_progress rows. The Today card and CLI now have no control on a due item.

What to build

Two additions on the existing candidate — no new ranking rule.

1. Age-aware nudge line. When an eligible due candidate is unrelated to every matchable plan and days_ago ≥ threshold, with the threshold derived from the constants (the smallest days_ago at which min(days_ago, 30) ≥ PLAN_RELATED_BIAS plus the plan-related candidate's own age — no second constant, so it moves with D-D), the candidate carries a nudge line: "Overdue 16 days and unrelated to your plan — do it, or retire it: studyloop review retire <topic> <concept>." Rendered where item 4's completion-review evidence lines render (CLI now beneath the action, Today card beside it, MCP payload as an additive key). A fact about age, not a proposal to rank differently.

2. Retire / snooze door for a due item, through the seam every surface uses:

Exposed like the plan lifecycle: a CLI verb (studyloop review retire|snooze), a Today-card control on the due item, one MCP tool beside record_study_progress (the mirror of record_topic_progress(confidence="resolved") for cards).

Constraints:

1. Learner-issued only — nothing retires or snoozes on the learner's behalf (the nudge proposes, the learner decides; same shape as item 4's consensual close).
2. History kept — excluded from the due set, never deleted; get_study_history and unverified_milestones still see it.
3. Plan-aware, not plan-bound — the nudge fires only for items unrelated to every matchable plan; a plan-related overdue item is the plan's business (item 4's closing review counts it).
4. Golden unchanged — the no-plan golden has no due items; the nudge is an additive key on a due candidate.
5. Threshold derived, not a new constant.

Out of scope

• REVIEW_INTERVALS and how days_ago is computed.
• Reranking — the overtaking-on-age behaviour is correct (row 2); what is missing is the explanation and the exit.
• Flashcard/quiz card_reviews (a different store with its own SM-2 scheduler in review_db.py); a retire door for those is a separate ticket if wanted.

Acceptance criteria

• RED: a due unrelated item past the derived threshold carries the nudge line on CLI, Today and MCP; a plan-related one does not; one below the threshold does not.
• RED: retire removes the row from the due set and from now candidates, keeps its history row, and record_progress on the same concept clears it; snooze excludes it until the date and not after.
• The migration lives in agent_session_tools.migrations and the clean-rebuild closure classifies the new column/state (the 2026-09-12 rebuild caught two tables the migrations did not create; do not repeat that).
• Golden now_plan_no_active.json byte-identical; test_docs_plan_integration_contract.py green with the doc sentence added to docs/study-plans.md "Plan-aware now" and docs/cli-reference.md.
• A D-16 rubric row: row 2's world advanced 16 days; the owner asked whether the nudge line is one they would act on.

Blocked by

• Nothing. Independent of D-D (the threshold derivation is written so it follows D-D's derived bias when that lands).

Definition of Done

• Full unit suite green; just lint; just typecheck.
• Council review of the RED and the arbitration before merge (one commit per accepted finding).
• Docs, CLI reference and MCP/CLI specs updated in the same change.