# Security Governance Claim Traceability Matrix This matrix links each top conclusion to evidence references, source URLs, quality appraisal, and upgrade blockers. It is generated from `security_governance/data/workbench.json`. ## Claims ### rc_01_molot_is_direct_mainline · L2_direct_public_evidence · blocked_or_boundary_claim - Claim: MOLOT 已可作为安全专题的直接论文主线，但它证明的是 Python/JavaScript 包 SAST 静态恶意代码检测，不证明 FCD C/C++ 表。 - Promotion status: `promoted` - Decision use: 顶部结论应从“候选模型”升级为“直接论文证据主线”。 - Evidence refs: `molot_system_card` - Source URLs: https://arxiv.org/abs/2606.07792 - Linked blockers: `block_fcd_pdg_table_upgrade`, `block_open_malicious_code_bench_artifact_upgrade`, `block_agent_gate_performance_upgrade` - Upgrade condition: Resolve linked blockers before any stronger wording is allowed. Evidence nodes: - `molot_system_card` — MOLOT System Card: Malicious Operational Logic Observation Transformer — authority=peer_reviewed, recency=current, directness=direct, allowed_use=current_claim, url=https://arxiv.org/abs/2606.07792 Blocked extrapolation: - FCD C/C++ performance table - LRI Agent Gate runtime/FPR - independent Open Malicious-Code Bench artifact URL ### rc_02_behavior_chain_becomes_detection_principle · L2_multi_paper_supported_principle · public_evidence_supported - Claim: 行为链检测已经从方法背景上移为安全专题的核心原则：单个 API/关键词不足以判断恶意目的，必须看 activity chain。 - Promotion status: `promoted` - Decision use: LRI gate taxonomy 应以 activity_chain 为一等字段，而不是 isolated keyword/API。 - Evidence refs: `molot_system_card`, `cerebro_behavior_sequence`, `npm_malicious_detection_benchmark` - Source URLs: https://arxiv.org/abs/2606.07792, https://arxiv.org/abs/2309.02637, https://arxiv.org/abs/2603.27549 - Linked blockers: none - Upgrade condition: No upgrade needed for current wording. Evidence nodes: - `molot_system_card` — MOLOT System Card: Malicious Operational Logic Observation Transformer — authority=peer_reviewed, recency=current, directness=direct, allowed_use=current_claim, url=https://arxiv.org/abs/2606.07792 - `cerebro_behavior_sequence` — Malicious Package Detection in NPM and PyPI using a Single Model of Malicious Behavior Sequence — authority=peer_reviewed, recency=aging, directness=background, allowed_use=background, url=https://arxiv.org/abs/2309.02637 - `npm_malicious_detection_benchmark` — Understanding NPM Malicious Package Detection: A Benchmark-Driven Empirical Analysis — authority=peer_reviewed, recency=current, directness=background, allowed_use=background, url=https://arxiv.org/abs/2603.27549 Blocked extrapolation: - binary-only detection - exact precision without benchmark table extraction ### rc_03_false_positive_budget_is_required · L2_direct_operational_constraint · blocked_or_boundary_claim - Claim: 安全门禁不能只追求召回率；误报预算和 block/ask_human 分流必须前置。 - Promotion status: `promoted` - Decision use: Agent Gate MVP 应默认区分 block、warn、ask_human，而非二元恶意/良性。 - Evidence refs: `python_malware_benchmark`, `npm_malicious_detection_benchmark`, `molot_system_card` - Source URLs: https://arxiv.org/abs/2209.13288, https://arxiv.org/abs/2603.27549, https://arxiv.org/abs/2606.07792 - Linked blockers: `block_agent_gate_performance_upgrade` - Upgrade condition: Resolve linked blockers before any stronger wording is allowed. Evidence nodes: - `python_malware_benchmark` — A Benchmark Comparison of Python Malware Detection Approaches — authority=peer_reviewed, recency=aging, directness=background, allowed_use=background, url=https://arxiv.org/abs/2209.13288 - `npm_malicious_detection_benchmark` — Understanding NPM Malicious Package Detection: A Benchmark-Driven Empirical Analysis — authority=peer_reviewed, recency=current, directness=background, allowed_use=background, url=https://arxiv.org/abs/2603.27549 - `molot_system_card` — MOLOT System Card: Malicious Operational Logic Observation Transformer — authority=peer_reviewed, recency=current, directness=direct, allowed_use=current_claim, url=https://arxiv.org/abs/2606.07792 Blocked extrapolation: - claiming LRI measured FPR before logs exist ### rc_04_explainability_is_review_evidence · L2_direct_plus_foundation_evidence · background_with_overclaim_risk - Claim: SHAP/解释不应停留在可视化，而应上移为 reviewer evidence：source_locations、activity_chain、reviewer_action 必须绑定。 - Promotion status: `promoted` - Decision use: reviewer_evidence_schema 应保持为工程化主产物。 - Evidence refs: `molot_system_card`, `shap_unified_approach` - Source URLs: https://arxiv.org/abs/2606.07792, https://arxiv.org/abs/1705.07874 - Linked blockers: none - Upgrade condition: No upgrade needed for current wording. Evidence nodes: - `molot_system_card` — MOLOT System Card: Malicious Operational Logic Observation Transformer — authority=peer_reviewed, recency=current, directness=direct, allowed_use=current_claim, url=https://arxiv.org/abs/2606.07792 - `shap_unified_approach` — A Unified Approach to Interpreting Model Predictions — authority=peer_reviewed, recency=aging, directness=background, allowed_use=background, url=https://arxiv.org/abs/1705.07874 Blocked extrapolation: - SHAP alone as causal proof - SHAP alone deciding block ### rc_05_graph_pdg_is_secondary_chain_not_molot_metric · L1_background_evidence_with_boundary · blocked_or_boundary_claim - Claim: PDG/Graph 证据可作为源码/片段相似检测的第二检测面，但不能上移为 MOLOT 性能或 FCD 表直接证据。 - Promotion status: `boundary_or_candidate` - Decision use: 工作台应保留 PDG/FCD 线，但所有性能数字必须阻断升级。 - Evidence refs: `pdg_holmes_graph_siamese`, `source_code_is_graph`, `fcd_pdg_performance_unresolved` - Source URLs: https://arxiv.org/abs/2011.11228, https://arxiv.org/abs/2312.16488, user_provided_json - Linked blockers: `block_fcd_pdg_table_upgrade` - Upgrade condition: Resolve linked blockers before any stronger wording is allowed. Evidence nodes: - `pdg_holmes_graph_siamese` — Modeling Functional Similarity in Source Code with Graph-Based Siamese Networks — authority=peer_reviewed, recency=aging, directness=background, allowed_use=background, url=https://arxiv.org/abs/2011.11228 - `source_code_is_graph` — Source Code is a Graph, Not a Sequence: A Cross-Lingual Perspective on Code Clone Detection — authority=peer_reviewed, recency=aging, directness=background, allowed_use=background, url=https://arxiv.org/abs/2312.16488 - `fcd_pdg_performance_unresolved` — FCD / PDG C/C++ Performance Table from User JSON — authority=user_input, recency=unverified, directness=unresolved, allowed_use=candidate_only, url=user_provided_json Blocked extrapolation: - verified C/C++ precision/recall/RMSE/FCD speed - equating graph clone papers with malicious-code detection metrics ### rc_06_agent_gate_is_design_candidate_not_verified_result · L0_to_L1_design_candidate · blocked_or_boundary_claim - Claim: LRI Agent Gate 已具备合同、schema 和 MVP 用例设计，但还不能上移为已验证性能结论。 - Promotion status: `boundary_or_candidate` - Decision use: 顶部结论应写成“工程设计候选”，下一步是跑 MVP audit log。 - Evidence refs: `molot_system_card`, `python_malware_benchmark` - Source URLs: https://arxiv.org/abs/2606.07792, https://arxiv.org/abs/2209.13288 - Linked blockers: `block_agent_gate_performance_upgrade` - Upgrade condition: Resolve linked blockers before any stronger wording is allowed. Evidence nodes: - `molot_system_card` — MOLOT System Card: Malicious Operational Logic Observation Transformer — authority=peer_reviewed, recency=current, directness=direct, allowed_use=current_claim, url=https://arxiv.org/abs/2606.07792 - `python_malware_benchmark` — A Benchmark Comparison of Python Malware Detection Approaches — authority=peer_reviewed, recency=aging, directness=background, allowed_use=background, url=https://arxiv.org/abs/2209.13288 Blocked extrapolation: - verified Agent Gate precision - verified latency/memory - verified false-positive budget ## Edges - `rc_01_molot_is_direct_mainline` --supported_by--> `molot_system_card` - `rc_01_molot_is_direct_mainline` --blocked_by--> `block_fcd_pdg_table_upgrade` - `rc_01_molot_is_direct_mainline` --blocked_by--> `block_open_malicious_code_bench_artifact_upgrade` - `rc_01_molot_is_direct_mainline` --blocked_by--> `block_agent_gate_performance_upgrade` - `rc_02_behavior_chain_becomes_detection_principle` --supported_by--> `molot_system_card` - `rc_02_behavior_chain_becomes_detection_principle` --supported_by--> `cerebro_behavior_sequence` - `rc_02_behavior_chain_becomes_detection_principle` --supported_by--> `npm_malicious_detection_benchmark` - `rc_03_false_positive_budget_is_required` --supported_by--> `python_malware_benchmark` - `rc_03_false_positive_budget_is_required` --supported_by--> `npm_malicious_detection_benchmark` - `rc_03_false_positive_budget_is_required` --supported_by--> `molot_system_card` - `rc_03_false_positive_budget_is_required` --blocked_by--> `block_agent_gate_performance_upgrade` - `rc_04_explainability_is_review_evidence` --supported_by--> `molot_system_card` - `rc_04_explainability_is_review_evidence` --supported_by--> `shap_unified_approach` - `rc_05_graph_pdg_is_secondary_chain_not_molot_metric` --supported_by--> `pdg_holmes_graph_siamese` - `rc_05_graph_pdg_is_secondary_chain_not_molot_metric` --supported_by--> `source_code_is_graph` - `rc_05_graph_pdg_is_secondary_chain_not_molot_metric` --supported_by--> `fcd_pdg_performance_unresolved` - `rc_05_graph_pdg_is_secondary_chain_not_molot_metric` --blocked_by--> `block_fcd_pdg_table_upgrade` - `rc_06_agent_gate_is_design_candidate_not_verified_result` --supported_by--> `molot_system_card` - `rc_06_agent_gate_is_design_candidate_not_verified_result` --supported_by--> `python_malware_benchmark` - `rc_06_agent_gate_is_design_candidate_not_verified_result` --blocked_by--> `block_agent_gate_performance_upgrade`