PART 01


PART 02



PART 03
平台 | 速度扰动 | 航向扰动 | 最大转弯率参考 |
RC-135V/W | 35 km/h | 4° | 1.2°/s |
F-16CM | 70 km/h | 9° | 4.5°/s |
F-22A | 80 km/h | 10° | 5.0°/s |
E-3G | 30 km/h | 3° | 1.0°/s |
KC-135R | 25 km/h | 3° | 0.8°/s |
P-8A | 40 km/h | 5° | 1.8°/s |
MQ-4C | 20 km/h | 2° | 0.7°/s |
C-130J | 25 km/h | 3° | 1.0°/s |
PART 04

威胁等级 | |
high | +0.18 |
medium | +0.10 |
low | -0.08 |
国别语义 | |
red | +0.08 |
unknown | +0.10 |
neutral | -0.12 |
blue | -0.18 |
支援 Link 类型 | |
support_enabler | +0.16 |
escort_or_counterair | +0.12 |
isr_collection_chain | +0.10 |
none | 0.00 |

PART 05
{"name": "predict_adiz_entry_probability","version": "0.2.0","input": ["AirTarget", "TrackPoint", "IntelReport", "SimilarTrackCase", "SupportRelationship", "AirspaceZone"],"output": ["AdizEntryProbability"],"riskLevel": "medium","skillBinding": {"skill": "adiz-entry-probability","version": "0.2.0","entrypoint": "skills/adiz-entry-probability/scripts/predict.py","progressiveDisclosure": true,"allowedEffect": "decision_support_only"}这个函数叫什么 输入哪些对象 输出什么对象 风险等级是什么 允许产生什么效果 绑定哪个Skill
什么时候应该使用 如何准备输入 如何运行脚本 参考哪些参数表 踩过哪些坑 如何解释输出
PART 06
work/ontology-runtime-mvp/skills/adiz-entry-probability work/ontology-runtime-mvp/skills/assess-threat
SKILL.md scripts/ references/ assets/ optional
---name: adiz-entry-probabilitydescription: Use when OpenClaw needs to estimate the probability that one or more air targets will enter the East China Sea ADIZ...version: 0.2.0ontology_function: predict_adiz_entry_probabilityallowed_effect: decision_support_only---
references/platform_params.md references/probability_fusion.md scripts/predict.py
PART 07
{"skill": {"name": "adiz-entry-probability","version": "0.2.0","loaded_levels": ["metadata", "SKILL.md", "scripts/predict.py"],"progressive_disclosure": true}PART 08
已进入 ADIZ 的目标直接 P=1.0 平台参数要区分 ISR、战斗机、加油机、预警机 保障关系用 logit prior,不用简单百分比加成
不能直接用经纬度差值算运动状态 不能提前显示未来预测航迹 单源情报不能过度抬高置信度 数据源不要直接连对象,要说明字段映射和对象属性
MissionPackage 任务编组对象 AirBaseActivity 机场活动对象 SensorEmission 电子辐射对象 CommandIntentHypothesis指挥意图假设对象 AirRouteCorridor 航路走廊对象 ReviewTask 人工复核任务对象
OpenClaw Memory 让智能体记住上下文 Skills 让函数能力积累经验 Dynamic Ontology 让业务语义空间持续扩展
PART 09

situation_feed recon_report sigint_report opensourcegeo historicalcasedb
AirTarget TrackPoint IntelReport AirspaceZone SimilarTrackCase SupportRelationship
assess_threat predictadizentry_probability@0.2.0 skill:assess-threat skill:adiz-entry-probability updatecesiumoverlay updateprobabilityoverlay
ThreatAssessment ApproachEvent AdizEntryProbability HumanReviewTask
Cesium Digital Earth OpenClaw Console Report/Audit View
PART 10
动态本体负责定义业务世界:有哪些对象、属性、Link、函数声明、动作和权限边界。 MCP Runtime 负责治理:匹配本体、准备 workspace、检查数据契约、调用函数、记录 lineage、拦截高风险动作。 OpenClaw Skills 负责能力实现:把函数背后的算法、脚本、参数表、经验文档沉淀下来,并支持渐进式加载。 OpenClaw 负责智能体编排:理解自然语言任务,选择本体,调用 MCP,解释结果,继续追问或发起复核。 Cesium 负责应用呈现:显示目标、航迹、预测、ADIZ、情报标记、概率卡片和本体图。
PART 11
这次为什么偏高 是平台参数问题 是历史相似案例问题 是情报权重问题 还是本体 Link先验问题
