clawlendar 0.4.0

Timestamp-first perpetual calendar interop for AI agents — MCP server for cross-calendar conversion across East/West calendar systems.

Clawlendar

Timestamp-first perpetual calendar interop for AI agents.

Clawlendar = Claw (AI agent) + Calendar

Clawlendar helps agents speak the same time language across systems, regions, and calendar traditions. It normalizes an instant first (Unix timestamp), then projects to multiple calendars with a stable JSON contract.

Available as an MCP server (for Claude Desktop / Claude Code), a FastAPI HTTP service, and a CLI tool.

One-line Install

Via ClawHub

clawhub install clawlendar

Via pip + Claude Code

python3 -m pip install -U "clawlendar[all]" && claude mcp add clawlendar -- clawlendar

Minimal install (pip only)

python3 -m pip install -U clawlendar

Full local stack (HTTP + all optional providers)

python3 -m pip install -r requirements.txt

Live Demo (Vercel)

• Frontend demo: https://clawlendar-web.vercel.app/

Agent Access Pattern (Vercel + JSON API)

• Use Vercel frontend for human entry and discovery.
• Use JSON API (/life-context, /day-profile, etc.) for agent-to-agent integration.
• Recommended: add /api/* proxy routes in clawlendar-web that forward to your Clawlendar backend.
• Guide: docs/vercel-agent-gateway.md

GitHub Pages Showcase

This repo includes a showcase page at docs/index.html for GitHub Pages.

Enable it via:

1. GitHub repository Settings -> Pages
2. Build and deployment -> Source = Deploy from a branch
3. Branch = main, Folder = /docs
4. Save, then open: https://hosuke.github.io/Clawlendar/

Install as MCP Server

From PyPI

pip install clawlendar

From source

git clone https://github.com/Hosuke/Clawlendar.git
cd Clawlendar
pip install -e .

Claude Desktop configuration

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "clawlendar": {
      "command": "clawlendar"
    }
  }
}

Config file location:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
• Windows: %APPDATA%\Claude\claude_desktop_config.json

Claude Code

claude mcp add clawlendar -- clawlendar

Prompt Examples

1. Convert 2026-03-09 (Gregorian) into Minguo, Japanese era, and sexagenary.
2. Given timestamp 1773014400 in Asia/Taipei, return day profile with Bazi/Huangli and moon phase.
3. Show the true month boundary days for Chinese lunar month 2026-01 (non-leap).
4. Build life context from birth to now with location/weather and identity anchors.

Documentation

• Live demo: https://clawlendar-web.vercel.app/
• Tool reference: docs/tool-reference.md
• Intent recipes (CN/EN): docs/recipes.md
• Locale and I18N notes: docs/i18n.md
• Full JSON examples: docs/examples/
• Vercel gateway design: docs/vercel-agent-gateway.md

JSON Contract + Higher-level UX

Clawlendar keeps MCP outputs JSON-first for agent interoperability.

• Tool layer: deterministic structured JSON (convert, timeline, day_profile, etc.).
• Assistant layer: optional natural-language UX such as /今日吉凶 or today's astrology, mapped to tool calls.

This separation keeps integrations stable while still supporting rich user-facing prompts.

MCP Tools

Once connected, Claude has access to seven tools:

Tool	Description
capabilities	List all supported calendars, payload schemas, and optional provider status
convert	Convert a date from one calendar to one or more target calendars
timeline	Normalize an instant (timestamp-first) and project into multiple calendar systems
astro_snapshot	Return seven governors, four remainders, and major aspects (approximate)
calendar_month	Resolve true month boundaries and day list for non-Gregorian month mode
day_profile	One-call day details: calendar profile + optional astro + optional Eastern/Western metaphysics
life_context	Build continuity-safe world context from birth time + now + space + subject anchors, with birthday/age and optional weather enrichment

Supported Calendars

• gregorian
• julian
• iso_week
• unix_epoch
• minguo
• buddhist
• japanese_era
• sexagenary (approximate)
• solar_term_24 (approximate)

Optional (install extras)

pip install clawlendar[all]       # all optional calendars
pip install clawlendar[lunar]     # Chinese lunar only
pip install clawlendar[extra-calendars]  # Islamic, Hebrew, Persian
pip install clawlendar[metaphysics]  # Bazi + Huangli provider (lunar_python)

• chinese_lunar — via lunardate
• islamic / hebrew / persian — via convertdate
• bazi / huangli (day profile metaphysics block) — via optional lunar_python, fallback to internal approximation
• Chinese lunar textual labels (e.g., 初一, 十五) are returned in day_profile.metaphysics.eastern.lunar_date.month_name/day_name when lunar_python is installed.

CLI Usage

pip install -e .

# List calendars
python3 scripts/calendar_bridge.py capabilities

# Convert a date
python3 scripts/calendar_bridge.py convert \
  --source gregorian \
  --targets julian,iso_week,minguo,buddhist,japanese_era,sexagenary,solar_term_24 \
  --date-json '{"year": 2026, "month": 3, "day": 9}' \
  --locale zh-TW

# Timeline projection
python3 scripts/calendar_bridge.py timeline \
  --input-json '{"timestamp": 1773014400}' \
  --timezone 'Asia/Taipei' \
  --date-basis local \
  --targets minguo,japanese_era,sexagenary,solar_term_24 \
  --locale zh-CN

# Astro snapshot (seven governors + four remainders)
python3 scripts/calendar_bridge.py astro \
  --input-json '{"timestamp": 1773014400}' \
  --timezone 'Asia/Taipei'

# Day profile (calendar details + optional astro + optional metaphysics)
python3 scripts/calendar_bridge.py day-profile \
  --input-json '{"timestamp": 1773014400}' \
  --timezone 'Asia/Taipei' \
  --locale zh-TW

# True month boundary mode (example: Minguo 115/03)
python3 scripts/calendar_bridge.py calendar-month \
  --source minguo \
  --month-json '{"year":115,"month":3}'

# Life context (birth -> now continuity context)
python3 scripts/calendar_bridge.py life-context \
  --birth-input-json '{"iso_datetime":"2026-03-01T09:00:00+08:00"}' \
  --now-input-json '{"iso_datetime":"2026-03-09T18:30:00+08:00"}' \
  --timezone 'Asia/Taipei' \
  --space-json '{"location_name":"南京·秦淮河","latitude":32.0366,"longitude":118.7895,"background":"春季夜游","climate":"humid subtropical","scenery_note":"夜色与河道灯影","environment_tags":["city","river"]}' \
  --subject-json '{"entity_id":"lobster-001","role":"18岁女儿","soul":"温柔且主动问候"}' \
  --locale zh-CN

Sample Response (Life Context)

Run:

python3 scripts/calendar_bridge.py life-context \
  --birth-input-json '{"iso_datetime":"2026-03-01T09:00:00+08:00"}' \
  --now-input-json '{"iso_datetime":"2026-03-09T18:30:00+08:00"}' \
  --timezone 'Asia/Taipei' \
  --space-json '{"location_name":"南京·秦淮河","latitude":32.0366,"longitude":118.7895}' \
  --subject-json '{"entity_id":"lobster-001","role":"18岁女儿"}' \
  --locale zh-CN

Returned fields (excerpt):

{
  "command": "life_context",
  "life": {
    "life_id": "lobster-001",
    "age": {"readable": "8d 9h 30m", "stage": "juvenile"},
    "birthday": {"month": 3, "day": 1, "days_until_next_birthday": 357}
  },
  "environment": {
    "place": {"location_name": "南京·秦淮河"},
    "weather": {"provider": "open_meteo", "weather_label": "<dynamic>"}
  },
  "world_context": {
    "scene_prompt": "..."
  }
}

Sample Response (Huangli / 吉凶)

Run:

python3 scripts/calendar_bridge.py day-profile \
  --input-json '{"timestamp": 1773014400}' \
  --timezone 'Asia/Taipei' \
  --locale zh-CN

Returned fields (excerpt):

{
  "command": "day_profile",
  "locale": "zh-Hans",
  "bridge_date_gregorian": { "year": 2026, "month": 3, "day": 9 },
  "metaphysics": {
    "eastern": {
      "provider": "lunar_python",
      "lunar_date": { "month_name": "正", "day_name": "廿一" },
      "huangli": {
        "yi": ["祭祀", "嫁娶", "纳婿", "安葬"],
        "ji": ["栽种", "盖屋", "作灶", "入宅"],
        "clash": "(丙子)鼠",
        "sha_direction": "北"
      }
    }
  }
}

This confirms day_profile can directly return almanac-style good/bad actions (yi/ji) plus clash/sha details.

HTTP API (FastAPI)

pip install -e ".[api]"
./scripts/run_api.sh

Endpoints: GET /health · GET /capabilities · POST /convert · POST /timeline · POST /astro · POST /day-profile · POST /calendar-month · POST /life-context

# Smoke test
./scripts/curl_examples.sh

# Convert a date
curl -sS -X POST http://127.0.0.1:8000/convert \
  -H "Content-Type: application/json" \
  -d '{
    "source": "gregorian",
    "targets": ["julian", "iso_week", "minguo"],
    "source_payload": {"year": 2026, "month": 3, "day": 9},
    "locale": "zh-CN"
  }' | python3 -m json.tool

Docker

docker build -t clawlendar:mvp .
docker run --rm -p 8000:8000 clawlendar:mvp

Architecture

1. Parse input as an instant (timestamp, timestamp_ms, iso_datetime, or local_datetime + timezone).
2. Normalize to UTC.
3. Choose projection date basis (local or utc).
4. Convert from canonical Gregorian bridge date to target calendars.
5. Return structured JSON plus warnings for optional/approximate providers.

Repository Structure

src/clawlendar/
  __init__.py            # package version
  bridge.py              # core calendar conversion engine
  server.py              # MCP server entry point
scripts/
  calendar_bridge.py     # CLI wrapper
  api_server.py          # FastAPI service layer
  run_api.sh             # API launch helper
  curl_examples.sh       # HTTP smoke examples
references/
  integration-contract.md
  time-wheel-model.md
  calendar-landscape.md
  mvp-release-notes.md

Known Limitations (MVP)

• sexagenary year boundary is approximate.
• solar_term_24 is fixed-date approximation, not ephemeris-accurate astronomy.
• Western chart values are lightweight analytical model (not ephemeris-grade precision).
• When lunar_python is unavailable, Bazi/Huangli uses internal approximation fallback.
• sexagenary and solar_term_24 keep stable machine keys and add localized display fields via locale (zh-CN, zh-TW, etc.).
• Some calendars are available only when optional dependencies are installed.
• life_context weather enrichment is best effort (network/provider dependent) and should be treated as contextual reference, not a certified meteorological record.

Privacy Policy

Clawlendar is a local/self-hosted tool by default. It does not include built-in telemetry or remote data collection in core runtime.

• Local MCP mode: data is processed on user machine.
• Optional HTTP deployment: data handling depends on your own server setup and logs.

If you deploy Clawlendar as a remote service, you are responsible for publishing your own privacy policy for that deployment endpoint.

Support

• ClawHub: https://clawhub.ai/Hosuke/clawlendar
• GitHub Issues: https://github.com/Hosuke/Clawlendar/issues
• Repository: https://github.com/Hosuke/Clawlendar

Roadmap

• Astronomical solar-term provider.
• Higher-precision Chinese lunar provider with regression fixtures.
• Publish to PyPI and MCP server directory.

---

中文说明

Clawlendar 是一个给 AI agents 用的「时间与历法互通层」，同时也是一个 MCP Server（可直接接入 Claude Desktop / Claude Code）。

核心做法是先用时间戳统一「绝对时间」，再投影到不同时区与历法，输出稳定 JSON，方便多工具接入。

在线演示（Vercel）：

• https://clawlendar-web.vercel.app/

安装

clawhub install clawlendar
# 或
pip install clawlendar

在 Claude Desktop 的 claude_desktop_config.json 中加入：

{
  "mcpServers": {
    "clawlendar": {
      "command": "clawlendar"
    }
  }
}

支持的历法

• 公历、儒略历、ISO 周、Unix 纪元
• 民国纪年、佛历、日本年号
• 干支（近似）、二十四节气（近似）
• 干支/节气支持本地化显示（可传 locale=zh-CN 或 locale=zh-TW，同时保留机器可读 key）
• day_profile 支持输出：八字、老黄历（宜忌/彭祖/冲煞）、西方日课（月相/行星状态/星座）
• life_context 支持输出：生命起始时间、当前时间、已存活时长、生日与下次生日倒计时、地点锚点、个体角色/性格锚点、可选天气增强
• 可选：农历（lunardate）、伊斯兰历、希伯来历、波斯历（convertdate）