# Security Governance Evidence Chain

This generated evidence chain follows the LRI review methodology standard. It links top conclusions to resource cards, direct paper evidence, public evidence snapshots, metadata blockers, and upgrade actions.

## Top refreshed conclusions

### rc_01_molot_is_direct_mainline · L2_direct_public_evidence

- 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。
- Evidence refs: molot_system_card
- URLs: https://arxiv.org/abs/2606.07792
- Decision use: 顶部结论应从“候选模型”升级为“直接论文证据主线”。
- 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

- 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。
- Evidence refs: molot_system_card, cerebro_behavior_sequence, npm_malicious_detection_benchmark
- URLs: https://arxiv.org/abs/2606.07792, https://arxiv.org/abs/2309.02637, https://arxiv.org/abs/2603.27549
- Decision use: LRI gate taxonomy 应以 activity_chain 为一等字段，而不是 isolated keyword/API。
- Blocked extrapolation: binary-only detection, exact precision without benchmark table extraction

### rc_03_false_positive_budget_is_required · L2_direct_operational_constraint

- 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。
- Evidence refs: python_malware_benchmark, npm_malicious_detection_benchmark, molot_system_card
- URLs: https://arxiv.org/abs/2209.13288, https://arxiv.org/abs/2603.27549, https://arxiv.org/abs/2606.07792
- Decision use: Agent Gate MVP 应默认区分 block、warn、ask_human，而非二元恶意/良性。
- Blocked extrapolation: claiming LRI measured FPR before logs exist

### rc_04_explainability_is_review_evidence · L2_direct_plus_foundation_evidence

- 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。
- Evidence refs: molot_system_card, shap_unified_approach
- URLs: https://arxiv.org/abs/2606.07792, https://arxiv.org/abs/1705.07874
- Decision use: reviewer_evidence_schema 应保持为工程化主产物。
- 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

- 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。
- Evidence refs: pdg_holmes_graph_siamese, source_code_is_graph, fcd_pdg_performance_unresolved
- URLs: https://arxiv.org/abs/2011.11228, https://arxiv.org/abs/2312.16488, user_provided_json
- Decision use: 工作台应保留 PDG/FCD 线，但所有性能数字必须阻断升级。
- 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

- 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。
- Evidence refs: molot_system_card, python_malware_benchmark
- URLs: https://arxiv.org/abs/2606.07792, https://arxiv.org/abs/2209.13288
- Decision use: 顶部结论应写成“工程设计候选”，下一步是跑 MVP audit log。
- Blocked extrapolation: verified Agent Gate precision, verified latency/memory, verified false-positive budget

## Evidence quality summary

- `molot_system_card`: authority=peer_reviewed, recency=current, directness=direct, replicability=public_url, allowed_use=current_claim
- `cerebro_behavior_sequence`: authority=peer_reviewed, recency=aging, directness=background, replicability=public_url, allowed_use=background
- `python_malware_benchmark`: authority=peer_reviewed, recency=aging, directness=background, replicability=public_url, allowed_use=background
- `npm_malicious_detection_benchmark`: authority=peer_reviewed, recency=current, directness=background, replicability=public_url, allowed_use=background
- `oss_malicious_package_analysis`: authority=peer_reviewed, recency=current, directness=background, replicability=public_url, allowed_use=background
- `backstabbers_knife_collection`: authority=peer_reviewed, recency=aging, directness=background, replicability=public_url, allowed_use=background
- `unveiling_malicious_logic`: authority=peer_reviewed, recency=current, directness=background, replicability=public_url, allowed_use=background
- `llm_pypi_malicious_detection`: authority=peer_reviewed, recency=current, directness=background, replicability=public_url, allowed_use=background
- `pdg_holmes_graph_siamese`: authority=peer_reviewed, recency=aging, directness=background, replicability=public_url, allowed_use=background
- `source_code_is_graph`: authority=peer_reviewed, recency=aging, directness=background, replicability=public_url, allowed_use=background
- `shap_unified_approach`: authority=peer_reviewed, recency=aging, directness=background, replicability=public_url, allowed_use=background
- `fcd_pdg_performance_unresolved`: authority=user_input, recency=unverified, directness=unresolved, replicability=not_reproducible, allowed_use=candidate_only

## Upgrade blockers

- `block_fcd_pdg_table_upgrade`: FCD/PDG C/C++ table precision/recall/RMSE/FCD speed is verified. → 待核验用户线索；可能属于源码/二进制相似片段检测方向。
- `block_open_malicious_code_bench_artifact_upgrade`: Open Malicious-Code Bench independent artifact URL is verified. → 论文声明释放 Open Malicious-Code Bench；独立 artifact URL 待补。
- `block_agent_gate_performance_upgrade`: LRI Agent Gate has verified precision/latency/FPR. → 工程设计候选，等待 MVP 测试验证。