Retrograde MCP
A Model Context Protocol server that provides real astronomical data (planetary positions, lunar phases, space weather) with astrological interpretations to help users make decisions about deployments, incident analysis, and scheduling.
README
Retrograde MCP
Because your CI/CD pipeline deserves to know that Mercury is in retrograde.
A Model Context Protocol server that surfaces real astronomical data — planetary positions from the NASA JPL DE421 ephemeris and geomagnetic activity from NOAA SWPC — wrapped in the kind of rigorous astrological commentary your incident reports have been missing.
No fake planetary positions. No hardcoded retrograde dates. Every result is computed from actual ephemeris data at query time. If Mars suddenly goes rogue out of schedule, you WILL know.
What it does
| Tool | Description |
|---|---|
get_planetary_status |
Motion status (direct / retrograde / stationary) for Mercury through Neptune, with ecliptic longitude, daily speed, and domain interpretation |
get_lunar_phase |
Current Moon phase with illumination and deployment recommendations |
get_space_weather |
Real-time Kp-index from NOAA SWPC; geomagnetic storm level |
get_cosmic_risk_score |
Composite risk score 0–100: retrograde planets + lunar phase + Kp-index |
should_i_do_it |
Yes/no astrological recommendation for any action (deploy, merge PR, force push, rewrite auth...) |
explain_incident |
Give it an incident description; receive a rigorous astrological root-cause analysis |
get_favorable_window |
Next calendar window when planetary and lunar conditions are relatively benign |
get_daily_briefing |
Morning cosmic standup: what's in the sky and what it means for your pipeline |
retrograde_history |
All retrograde periods for any planet over the last N years — overlay against your deploy log |
Analysis methodology
Planetary retrograde detection uses geocentric ecliptic longitude computed from the NASA JPL DE421 ephemeris (downloaded automatically via Skyfield). A planet is classified as:
- Retrograde when its ecliptic longitude is decreasing (speed < 0) and the absolute rate exceeds the planet's stationary threshold
- Stationary when the absolute rate of change falls below a per-planet threshold (the brief reversal period before/after retrograde) — thresholds follow professional-ephemeris convention: Mercury 0.083°/day, Venus 0.050°/day, Mars 0.025°/day, Jupiter/Saturn 0.017°/day, Uranus 0.006°/day, Neptune 0.003°/day
- Direct when its ecliptic longitude is increasing and the absolute rate exceeds the planet's stationary threshold
Lunar phase is computed from the angular separation between the Moon and the Sun in ecliptic coordinates, also from DE421.
Space weather (Kp-index) is fetched in real time from services.swpc.noaa.gov. The server tries the 1-minute feed first, falls back to the 3-hour consolidated product.
Cosmic risk score weights:
- Mercury and Mars retrograde: +20 each (critical infra/communication planets)
- Other retrograde planets: +10 each
- Stationary planets: +5 each
- Lunar phase: −10 to +20
- Kp-index: −5 to +50
Installation
Prerequisites
- Python 3.10+
- ~20 MB disk for the JPL ephemeris file (downloaded automatically on first run)
- Internet access for NOAA real-time Kp (gracefully degraded if unavailable)
With uv (recommended)
uvx retrograde-mcp
With pip
pip install retrograde-mcp
retrograde-mcp
From source
git clone https://github.com/Griboedow/Retrograde-MCP
cd Retrograde-MCP
pip install -e .
retrograde-mcp
Claude Desktop configuration
Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or
%APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"retrograde": {
"command": "retrograde-mcp"
}
}
}
Or with uvx:
{
"mcpServers": {
"retrograde": {
"command": "uvx",
"args": ["retrograde-mcp"]
}
}
}
Ephemeris cache location
The JPL DE421 file is cached to ~/.retrograde-mcp/ by default.
Override with the RETROGRADE_CACHE_DIR environment variable:
{
"mcpServers": {
"retrograde": {
"command": "retrograde-mcp",
"env": {
"RETROGRADE_CACHE_DIR": "/tmp/retrograde-cache"
}
}
}
}
Usage examples
"Should I deploy right now?"
should_i_do_itget_cosmic_risk_scoreCosmic Risk Score: 🟠 ELEVATED (45/100)
Blocking factors:
- Geomagnetic storm (G1) in progress (Kp=5.7) — mental clarity across the team may be compromised
Caution factors:
- Full Moon 🌕 — maximum lunar intensity, not favorable for deployments
In your favor: Mercury, Mars, and Venus are all direct and supportive.
Recommendation: Hold off. Use
get_favorable_windowto find a better moment.
"Why did our API go down?"
explain_incident("API gateway returned 504s for 30 minutes starting at 14:00 UTC")All planets are currently direct. The incident represents a triumph of purely human error, unassisted by celestial interference.
Contributing factors:
- Uranus (stationary at 58.51°) — disruption of unexpected outages, paradigm shifts, zero-day exploits
- Neptune (stationary at 1.94°) — unclear requirements, scope creep, ambient confusion
- Full Moon (98% illumination) — amplified cosmic instability
- Kp-index: 5.7 — G1 geomagnetic storm in progress
Remediation:
- Document the incident with planetary positions recorded above
- Schedule the post-mortem during a Mercury-direct period (you're clear — Mercury went direct on March 20)
- Implement fixes during a waxing Moon phase for maximum cosmic support
- Use
retrograde_historyto check if similar incidents correlate with the same planetary configurations
"When can I safely ship next?"
get_favorable_window·get_cosmic_risk_scoreNext favorable window: April 19–29, 2026 — 11 days of cosmic cooperation with zero retrograde planets.
Today (April 3) scores 🟠 ELEVATED (45/100):
- Uranus & Neptune stationary (+10)
- Full Moon (+20)
- G1 geomagnetic storm, Kp=5.7 (+15)
No planets are retrograde, which is good — but the Full Moon and active geomagnetic storm push the risk up. If the release is ready and well-tested with solid rollback plans, today is survivable. If you can wait, April 19 is when the cosmos fully clear out for you.
"Give me the morning briefing"
get_daily_briefingCosmic Risk: MODERATE (27/100)
Stationary: Uranus, Neptune. Direct: Mercury, Venus, Mars, Jupiter, Saturn. Full Moon — 98% illumination. Kp-index: 1.7 — Quiet.
Proceed mindfully. Conditions are workable. Prioritize well-tested releases over experiments. Keep rollback procedures fresh.
"What was Mercury doing during our outages last months?"
retrograde_historyMercury was retrograde from February 26 to March 20, 2026 — covering most of last month. That's 22 days where Mercury's apparent motion was reversed, which, according to the ephemeris data, means your March outages had a cosmically valid root cause.
The two other retrograde periods in the past year were: Jul 18 – Aug 11, 2025 (24 days) Nov 9 – Nov 29, 2025 (19 days)
If your outages clustered in early-to-mid March, they fell squarely within retrograde window #3. Cross-reference your incident log timestamps against Feb 26 – Mar 20 and you may find a suspicious overlap.
Development
pip install -e ".[dev]"
pytest
Data sources
| Source | What it provides | URL |
|---|---|---|
| NASA JPL DE421 | Planetary and lunar positions | https://ssd.jpl.nasa.gov/ftp/eph/planets/bsp/de421.bsp |
| NOAA SWPC 1-min Kp | Near-real-time Kp-index | https://services.swpc.noaa.gov/json/planetary_k_index_1m.json |
| NOAA SWPC 3-hour Kp | Consolidated Kp-index (fallback) | https://services.swpc.noaa.gov/products/noaa-planetary-k-index.json |
Disclaimer
Planetary positions are computed from real astronomical data. The causal relationship between Mercury's ecliptic velocity and your deployment success rate has not been peer-reviewed. The author accepts no liability for outages, regressions, or existential crises arising from following or ignoring this advice.
That said: Mercury has been retrograde during a statistically suspicious number of incidents. We're just saying.
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。