Paths: File paths (shared/, references/, ../ln-*) are relative to skills repo root.

Type: L3 Worker Category: 3XX Planning

Review Repair Worker

Mandatory Read

MANDATORY READ: Load shared/references/evaluation_worker_runtime_contract.md, shared/references/evaluation_summary_contract.md, shared/references/cleanup_evidence_contract.md MANDATORY READ: Load ../ln-310-multi-agent-validator/references/context_review_pipeline.md, ../ln-310-multi-agent-validator/references/penalty_points.md MANDATORY READ: Load ../ln-310-multi-agent-validator/references/structural_validation.md, ../ln-310-multi-agent-validator/references/standards_validation.md, ../ln-310-multi-agent-validator/references/solution_validation.md, ../ln-310-multi-agent-validator/references/workflow_validation.md, ../ln-310-multi-agent-validator/references/quality_validation.md, ../ln-310-multi-agent-validator/references/dependency_validation.md, ../ln-310-multi-agent-validator/references/cross_reference_validation.md, ../ln-310-multi-agent-validator/references/risk_validation.md, ../ln-310-multi-agent-validator/references/premortem_validation.md, ../ln-310-multi-agent-validator/references/traceability_validation.md

Purpose

• apply accepted low-risk repairs using the 11-group systematic validation structure
• keep repairs separate from aggregation and approval
• record cleanup evidence if background tools are launched

Mode Gate

• mode=story: full 11-group repair using findings from ln-312
• mode=plan_review|context: apply accepted corrections only (max 5 corrections per ../ln-310-multi-agent-validator/references/context_review_pipeline.md Compare & Correct Safety Rules)

11-Group Validation Structure (mode=story)

The checklists below are mandatory inputs loaded above. Use the table only as the routing map for applying them.

#	Group	Criteria	Checklist
1	Structural	#1-#4, #23-#24	../ln-310-multi-agent-validator/references/structural_validation.md
2	Standards	#5	../ln-310-multi-agent-validator/references/standards_validation.md
3	Solution	#6, #21, #28	../ln-310-multi-agent-validator/references/solution_validation.md
4	Workflow	#7-#13	../ln-310-multi-agent-validator/references/workflow_validation.md
5	Quality	#14-#15	../ln-310-multi-agent-validator/references/quality_validation.md
6	Dependencies	#18-#19/#19b	../ln-310-multi-agent-validator/references/dependency_validation.md
7	Cross-Reference	#25-#26	../ln-310-multi-agent-validator/references/cross_reference_validation.md
8	Risk	#20	../ln-310-multi-agent-validator/references/risk_validation.md
9	Pre-mortem	#27	../ln-310-multi-agent-validator/references/premortem_validation.md
10	Verification	#22	../ln-310-multi-agent-validator/references/traceability_validation.md
11	Traceability	#16-#17, #17b-#17c	../ln-310-multi-agent-validator/references/traceability_validation.md

Repair Rules

• Zero out penalty points only when the defect is actually repaired (not just acknowledged).
• Use FLAGGED only when human judgment is required and auto-fix cannot safely continue.
• Maximum penalty per ../ln-310-multi-agent-validator/references/penalty_points.md (do not hardcode).
• Test strategy section may exist but remain empty.
• Apply auto-fix actions exactly as specified in each checklist's "Auto-fix actions" column.

Runtime

Runtime family:

• evaluation-worker-runtime

Required manifest fields:

• identifier
• phase_order
• summary_kind=review-repair
• operation=repair

Recommended phase_order:

1. PHASE_0_CONFIG
2. PHASE_1_LOAD_FINDINGS
3. PHASE_2_GROUP_STRUCTURAL (#1-#4, #23-#24)
4. PHASE_3_GROUP_STANDARDS_SOLUTION (#5, #6, #21, #28)
5. PHASE_4_GROUP_WORKFLOW_QUALITY (#7-#15)
6. PHASE_5_GROUP_DEPS_XREF (#18-#19/#19b, #25-#26)
7. PHASE_6_GROUP_RISK_PREMORTEM (#20, #27)
8. PHASE_7_GROUP_VERIFICATION_TRACE (#22, #16-#17, #17b-#17c)
9. PHASE_8_VERIFY_LOCAL_RESULT
10. PHASE_9_WRITE_SUMMARY
11. PHASE_10_SELF_CHECK

Summary

Emit summary_kind=review-repair.

Payload must include:

• worker=ln-314
• status
• operation=repair
• warnings

Prefer these fields when available:

• penalty_before (from ln-312 findings)
• penalty_after (after repairs)
• flagged_items (list of items requiring human judgment)
• coverage_summary (AC coverage percentage)
• groups_processed (count)

Definition of Done

• Findings loaded from ln-312
• All 11 groups processed (mode=story) or accepted corrections applied (mode=plan_review|context)
• Penalty before/after tracked
• FLAGGED items recorded
• Local verification completed
• Cleanup evidence recorded when needed
• review-repair summary written
• Self-check passed

Version: 1.0.0 Last Updated: 2026-04-10