Saisca – offline supply chain risk analyzer (Excel/CSV → insights)

Saisca, a new open-source supply chain risk analysis tool, has been released on GitHub, allowing supply chain managers to import Excel or CSV files and receive risk insights, reasoning trails, and actionable recommendations entirely offline with no cloud dependency. The tool generates visual outputs including risk trend charts, propagation timelines, and high-risk node tables, with each high-risk node accompanied by a six-step reasoning trail and structured action recommendations. Saisca is now free and open under the MIT License, deployable locally via a downloaded `.dmg` file or through a cloned repository.

A locally-deployable supply chain risk analysis tool for supply chain managers. Import Excel/CSV, get risk insights, reasoning trails, and actionable recommendations — all offline, no cloud dependency. | Output | Answers the Question | |---|---| Risk Trend Chart | Is risk worsening or improving over time? | Risk Distribution Chart | Where is risk concentrated across the supply chain? | Propagation Timeline | Where does risk come from and spread to? | High-Risk Node Table | Which nodes need immediate action, in what order? | Data Confidence | How reliable is this analysis? What data is missing? | Domain Insights | Bullwhip effect, VMI, QR patterns detected? | Each high-risk node includes a full 6-step reasoning trail , risk cause details trigger metrics + threshold comparisons , and structured action recommendations replenish / reroute / switch supplier / adjust logistics / investigate / monitor . Download the latest .dmg from GitHub Releases https://github.com/cayincoorts-hue/saisika/releases → drag to Applications → launch Saisca. First launch: right-click → Open. Now free and open — no activation code required. git clone https://github.com/cayincoorts-hue/saisika.git cd saisika pip install -r requirements.txt python run.py Open http://localhost:8000 demo data/demo scenario/ contains a purpose-built 10-node supply chain simulation 26 weeks × 5 metrics : | File | Description | |---|---| Sales Order.csv | Downstream demand long format: date, node id, value | Production.csv | Factory output | Delivery To Distributor.csv | Distribution center deliveries | Factory Issue.csv | Supplier shipments | Inventory.csv | Node inventory levels S002 sharp drop in last 10 weeks | Nodes.csv | Node attributes name, type, tier, region | Edges.csv | Supply chain relationships risk links marked | Node Types.csv | Node type descriptions | README.md | Scenario design notes | Engineered risk characteristics: - S002 East China Parts Supplier → inventory crash + delivery disruption → high-risk node - P001 Shenzhen Factory → upstream volatility amplification → bullwhip effect - D001 South China DC → volatility significantly lower than peers → VMI pattern - R002 Shanghai Retailer → high-frequency small-batch stable demand → QR pattern User uploads Excel/CSV ↓ excel adapter Read file → detect role fact/node/edge/metadata ↓ field mapper Wide-table melt → column mapping → tri-state identification ↓ data merger Cross-scenario merge → mark duplicate measurement bases ↓ graph builder Build supply chain network → infer tier hierarchy ↓ risk engine Compute risk scores → annotate risk causes ↓ decision engine Classify actions → generate action type + justification ↓ analysis engine Assemble 6 result objects including domain insights ↓ prompt builder Generate text conclusions reads annotations, never raw numbers ↓ result exporter Output JSON + HTML report | Layer | Technology | |---|---| | Frontend | React + TypeScript + Vite + ECharts + GSAP | | Backend | Python + FastAPI + pandas + openpyxl | | Graph Layout | Three.js + react-force-graph-3d | | LLM optional | Ollama explanation layer only, not risk engine | | Packaging | PyInstaller + Electron + electron-builder | | Theory | SCOR / Bullwhip Effect / VMI / QR / SC-BSC | The system accepts two fact-table formats: Long format recommended : date, node id, value date,node id,value 2026-01-05,S001,99.3 Wide format : rows as dates, columns as node codes Date,SOS008L02P,SOS005L04P 2026-01-05,1355.0,890.2 Node tables need: node id, node name, node type . Edge tables need: source, target . - All data stored locally in data/ directory - No internet connection, no telemetry, no cloud dependency - Backend compiled to binary for code logic protection MIT License. See LICENSE /cayincoorts-hue/saisika/blob/main/LICENSE for details. 中文说明 面向供应链管理者，可本地部署、支持 Excel/CSV 导入的风险分析桌面工具。不依赖云端，数据不出本地。 | 结果 | 回答的问题 | |---|---| | 风险趋势图 | 风险是在加剧还是缓解？ | | 风险分布图 | 整体风险集中在哪一层？ | | 风险传播时序图 | 风险从哪里来、往哪里扩散？ | | 高风险节点表 | 哪些节点需要立即处理、按什么顺序？ | | 数据可信度 | 结论有多可信、缺了什么数据？ | | 供应链领域洞察 | 是否存在牛鞭效应、VMI、QR 等经典模式？ | 从 GitHub Releases https://github.com/cayincoorts-hue/saisika/releases 下载 .dmg → 拖入 Applications → 右键打开。 现已免费开放，无需激活码。 demo data/demo scenario/ 含 10 节点供应链模拟数据（26 周 × 5 指标），刻意设计了牛鞭效应、VMI、QR 等经典风险特征。 全部数据存本地 data/ 目录。不联网，无遥测，无云端依赖。 MIT License。详见 LICENSE /cayincoorts-hue/saisika/blob/main/LICENSE 。