shiryan-mcp-server
Enables AI agents to access SHIRYAN's change detection, geofencing, and alert synthesis for monitoring pipeline and critical infrastructure corridors in Saudi Arabia via MCP stdio tools.
README
🛰️ SHIRYAN (شريان) — The Energy-Artery Sentinel
Sovereign, onboard-AI infrastructure-integrity intelligence for Saudi Arabia. "We see the excavator before the rupture." — built for SARI 2 (Saudi Space Agency).
SHIRYAN watches the Kingdom's pipeline & critical-infrastructure corridors, detects third-party intrusion / geohazards / leaks from before/after imagery, and downlinks the decision (a small alert), not the image — delivered to operators and to AI agents via an MCP server, and shown live in the SHIRYAN Watch demo.
This repository is the ground-side product, MCP server, and pitch demo. The flight satellite is built later with SSA. See DESIGN.md for the spec and scope.
Repo layout
shiryan/
├─ packages/
│ ├─ core/ @shiryan/core — change detection, ROW geofencing, alert synthesis (pure TS, tested)
│ ├─ data/ @shiryan/data — corridors, sample scenes, Sentinel-2 converters
│ └─ mcp-server/ @shiryan/mcp-server — SHIRYAN intelligence as MCP tools (stdio, 4 tools)
├─ apps/
│ └─ web/ SHIRYAN Watch — MapLibre live demo (builds offline)
├─ scripts/ fetch-sentinel.ts — optional live Copernicus fetch (env creds; graceful fallback)
├─ data/ sample corridors + cached Sentinel-2 tiles
├─ docs/ architecture, pitch, Obsidian links
└─ DESIGN.md README.md STATE.md PROGRESS.md AGENT.md AGENTS.md memory.md LOG.md NOTES.md DECISIONS.md
Quick start
npm install # installs workspace dev deps (typescript, @types/node)
npm run verify # typecheck + tests (the gate — must be green)
npm run typecheck # types only
npm test # tests only (node --test, runs TypeScript natively)
Requires Node ≥ 22 (developed on Node 24; uses native TypeScript type-stripping, no build step for tests).
Quickstart
bash scripts/quickstart.sh
Runs an end-to-end offline SHIRYAN flow (load corridor → detect change → emit alerts → persist & read back history) and prints the result. One command, no network.
Status
| Component | State |
|---|---|
@shiryan/core |
✅ working + tested |
@shiryan/data |
✅ corridors + sample scene + geojson |
@shiryan/mcp-server |
✅ stdio MCP, 4 tools, tested over stdio |
apps/web (SHIRYAN Watch) |
✅ MapLibre demo, builds offline (npm run build -w @shiryan/web) |
| Real Sentinel-2 fetch | ⬜ planned (P4) |
| Full backlog: PROGRESS.md. |
Obsidian
Research, strategy, scoring, and the winning dossiers live in the Obsidian vault:
../Sarri 2 Vault/00_SARI2_Idea_Engine/ (flagship dossier: Round2/Model-1/Ideas/flagship.md).
This repo is the build; the vault is the brain. See memory.md.
CI
One-command gate for any runner: bash scripts/verify.sh (typecheck + lint + test + build, offline). Per-package coverage: bash scripts/coverage.sh. See AGENT.md.
Built autonomously
Developed via an overnight night-loop: fresh context each pass, filesystem-as-memory (DESIGN+README+ledger), one verified task per worker, green-gate-only commits. See AGENTS.md.
推荐服务器
Baidu Map
百度地图核心API现已全面兼容MCP协议,是国内首家兼容MCP协议的地图服务商。
Playwright MCP Server
一个模型上下文协议服务器,它使大型语言模型能够通过结构化的可访问性快照与网页进行交互,而无需视觉模型或屏幕截图。
Magic Component Platform (MCP)
一个由人工智能驱动的工具,可以从自然语言描述生成现代化的用户界面组件,并与流行的集成开发环境(IDE)集成,从而简化用户界面开发流程。
Audiense Insights MCP Server
通过模型上下文协议启用与 Audiense Insights 账户的交互,从而促进营销洞察和受众数据的提取和分析,包括人口统计信息、行为和影响者互动。
VeyraX
一个单一的 MCP 工具,连接你所有喜爱的工具:Gmail、日历以及其他 40 多个工具。
graphlit-mcp-server
模型上下文协议 (MCP) 服务器实现了 MCP 客户端与 Graphlit 服务之间的集成。 除了网络爬取之外,还可以将任何内容(从 Slack 到 Gmail 再到播客订阅源)导入到 Graphlit 项目中,然后从 MCP 客户端检索相关内容。
Kagi MCP Server
一个 MCP 服务器,集成了 Kagi 搜索功能和 Claude AI,使 Claude 能够在回答需要最新信息的问题时执行实时网络搜索。
e2b-mcp-server
使用 MCP 通过 e2b 运行代码。
Neon MCP Server
用于与 Neon 管理 API 和数据库交互的 MCP 服务器
Exa MCP Server
模型上下文协议(MCP)服务器允许像 Claude 这样的 AI 助手使用 Exa AI 搜索 API 进行网络搜索。这种设置允许 AI 模型以安全和受控的方式获取实时的网络信息。