fuse_spatial_evidence
Use this skill when the user wants to combine hotspot, anomaly, profile, forecast, and event-impact outputs into a unified ranked regional evidence view.
Файлы6
Открыть в GitHubSKILL.mdskills/custom/spatiotemporal_trajectory/fuse_spatial_evidence/SKILL.md
GitHubfuse_spatial_evidence
Use this Skill after any combination of $analyze_region_heat, $detect_flow_anomaly, $forecast_region_flow, $profile_urban_region, and $analyze_event_impact.
Command
cd /mnt/skills/custom/fuse_spatial_evidence
python3 scripts/fuse_spatial_evidence.py \
--input /path/to/flow_anomalies.jsonl \
--input /path/to/region_profiles.jsonl \
--input /path/to/event_impacts.jsonl \
--output-dir /path/to/output
Outputs
fused_evidence.jsonlevidence_details.jsonlsummary.json
Algorithm
- Evidence standardization
- Region/time alignment
- Cross-source score accumulation
- Dominant signal summarization
Next Skills
Use fused evidence as the final evidence layer for reports, briefings, or LLM-grounded answer generation.