菜单
Run the arktrace shadow fleet screening pipeline for a region. Use when scoring vessels, refreshing the watchlist, or running the full pipeline end-to-end.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Deploy arktrace to Cloudflare Pages and configure R2, CI, and environment variables. Use when setting up a new deployment or updating production configuration.
Download the arktrace demo dataset from R2 for local development. Use when you need sample data without running the full pipeline.
Configure the LLM provider for arktrace (OpenAI, local Ollama, or llama.cpp). Use when setting up a new environment or switching LLM providers.
Run the arktrace analyst dashboard locally. Use when developing the React frontend or verifying UI changes.
Run arktrace tests — pipeline unit/integration tests and frontend Vitest tests. Use before committing or when CI fails.
| name | arktrace-run-pipeline |
| description | Run the arktrace shadow fleet screening pipeline for a region. Use when scoring vessels, refreshing the watchlist, or running the full pipeline end-to-end. |
| license | Apache-2.0 |
| compatibility | Requires uv, DuckDB, .env with API keys |
| metadata | {"repo":"arktrace"} |
uv run python scripts/run_pipeline.py
Prompts for region selection and handles all flags automatically.
uv run python scripts/run_pipeline.py --region singapore --non-interactive
uv run python scripts/run_pipeline.py --region japan --non-interactive
Available regions: singapore, japan, middleeast, europe, persiangulf, gulfofguinea, gulfofaden, gulfofmexico, blacksea
uv run python -m pipeline.src.ingest.schema
uv run python -m pipeline.src.ingest.marine_cadastre
uv run python -m pipeline.src.ingest.sanctions
uv run python -m pipeline.src.ingest.vessel_registry
uv run python -m pipeline.src.features.ais_behavior
uv run python -m pipeline.src.features.identity
uv run python -m pipeline.src.features.ownership_graph
uv run python -m pipeline.src.features.trade_mismatch
uv run python -m pipeline.src.score.mpol_baseline
uv run python -m pipeline.src.score.anomaly
uv run python -m pipeline.src.score.causal_sanction
uv run python -m pipeline.src.score.composite
uv run python -m pipeline.src.score.watchlist
uv run python -m pipeline.src.ingest.eo_gfw --bbox 95,1,110,6 --days 30
# Without token:
uv run python -m pipeline.src.ingest.eo_gfw --csv data/raw/eo_detections_sample.csv