auseklis
Astrology MCP server that computes natal charts, transits, synastry, progressions, returns, eclipses, retrogrades, and moon phases from a real ephemeris, enabling AI agents to provide accurate astrological calculations without hallucination.
README
auseklis
Astrology MCP server — natal charts, transits, synastry, progressions, returns, eclipses, retrogrades, moon phases. Computed from a real ephemeris, so AI agents stop hallucinating planet positions.
Named after the Latvian morning star. MIT-licensed with no AGPL ephemeris data — see Licensing.
Tools
| Tool | What it does |
|---|---|
get_planet_position |
Position of one body/point at a moment (sign, degree, speed, retrograde) |
compute_natal_chart |
Full birth chart: 13 points, houses, angles, aspects |
compute_transits |
Aspects from the current (or any) sky to a natal chart |
compute_progressions |
Secondary progressions (day-for-a-year) |
compute_synastry |
Cross-chart aspects between two people |
compute_composite_chart |
Midpoint composite chart of a relationship |
find_returns |
Solar/lunar/planetary returns (exact moments) |
get_moon_phase |
Phase, illumination, Moon sign, next four quarters |
find_eclipses |
Lunar/solar eclipses with signs, incl. local visibility |
find_retrograde_periods |
Station retrograde/direct moments for any planet |
find_sign_ingresses |
When a body changes signs (equinoxes, Saturn ingresses, …) |
find_aspect_times |
Exact moment a transit perfects ("when does Saturn square my Sun?") |
Plus two prompts (natal_chart_reading, current_sky_report) and a glossary resource (auseklis://glossary).
Features: local birth times with IANA timezones (full historical DST handling) · tropical and sidereal (Lahiri, Fagan/Bradley) zodiacs · whole-sign, equal, Porphyry, and Placidus houses · mean lunar nodes and Black Moon Lilith.
Installation
Claude Code
claude mcp add auseklis -- npx -y auseklis
Claude Desktop / any MCP client (stdio)
{
"mcpServers": {
"auseklis": {
"command": "npx",
"args": ["-y", "auseklis"]
}
}
}
No API keys, no configuration — the ephemeris is computed locally.
Desktop Extension
Download auseklis.mcpb from the releases page and double-click to install in Claude Desktop. Or build it yourself: npm run bundle.
Remote (self-hosted)
The same server runs as a Cloudflare Worker speaking Streamable HTTP. Deploy it to your own account:
npm run deploy # wrangler deploy
claude mcp add --transport http auseklis https://auseklis.<your-subdomain>.workers.dev/mcp
Set the MCP_SHARED_SECRET secret to require a bearer token.
Example questions to ask
- "Compute my natal chart — born 15 March 1990, 15:45 in Riga."
- "What's transiting my Sun this month?"
- "When exactly is my Saturn return?"
- "Synastry between me and my partner?" (two birth date/times)
- "When is Mercury retrograde in 2027, and in which signs?"
- "Is tonight's full moon visible as an eclipse from here?"
The model handles place-name → coordinates; the server handles local-time → UTC via the IANA timezone database.
Accuracy
Positions come from astronomy-engine (VSOP87 + NOVAS C 3.1): ±1 arcminute for 1700–2200, far below the 1° resolution astrological interpretation uses. Event searches (stations, ingresses, returns, quarters) are refined to ~1 second of time. Verified in CI against published eclipse dates, the 2026 equinox, NOVAS Sun positions, and an independent Placidus implementation.
Architecture
src/
├── ephemeris/ Astrology core — backend-agnostic
│ ├── engine.ts EphemerisBackend interface + astronomy-engine adapter (the swap seam)
│ ├── index.ts Charts, aspects, synastry, composite, progressions
│ ├── events.ts Time searches: returns, stations, ingresses, aspect times, moon phases
│ ├── eclipses.ts Eclipse searches with astrological context
│ ├── houses.ts Whole-sign, equal, Porphyry, Placidus (semi-arc solver)
│ ├── points.ts Mean lunar nodes, Black Moon Lilith
│ ├── sidereal.ts Ayanamsa (Lahiri, Fagan/Bradley)
│ └── time.ts IANA timezone → UTC conversion (no dependencies, uses Intl)
├── mcp/ Tool/prompt/resource definitions on @modelcontextprotocol/sdk
├── stdio.ts Local entry — `npx auseklis`
└── index.ts Remote entry — Cloudflare Worker, Streamable HTTP via @hono/mcp
The EphemerisBackend interface in engine.ts is the deliberate swap seam: a future Rust/WASM clean-room ephemeris only needs to reimplement that one interface.
Development
npm install
npm run typecheck # strict TS
npm test # 20-check smoke suite (ephemeris references + MCP end-to-end)
npm run build # emit dist/
npm run dev # local Cloudflare Worker on :8787
npx @modelcontextprotocol/inspector node dist/stdio.js # poke tools interactively
See docs/tools.md for the full tool reference and docs/architecture.md for design notes.
Licensing
MIT. This project deliberately avoids the Swiss Ephemeris (.se1/.se2 data files and the sweph bindings): those are AGPL-licensed, which would impose AGPL obligations on any network service built on them. Everything here is computed from MIT-licensed code with no external data files — safe to embed, fork, and deploy commercially. Details in NOTICE.
The trade-off: no Chiron or asteroids (they need ephemeris data files). They are on the roadmap via public-domain JPL-derived data.
Roadmap
- v1: Rust/WASM clean-room ephemeris backend behind the same
EphemerisBackendseam - Chiron + major asteroids from public-domain JPL data
- Koch houses, declination/parallel aspects
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。