CorporateTravel Dispatch MCP
MCP server exposing CS Executive Services dispatch platform and airplanes.live flight tracking as portable, agent-agnostic tools.
README
corporatetravel-dispatch-mcp
Right repo? You want
corporatetravel-dispatch-mcp(this one) for the MCP server. The dispatch platform source lives atcorporatetraveldc-dispatch. The public mirror is atctdi-dispatch.
MCP server exposing the CS Executive Services dispatch platform and airplanes.live flight tracking as portable, agent-agnostic tools. Works with any MCP-compatible agent: Claude Code, Cline, Cursor, Zed, Windsurf, and others via mcpo → Open WebUI.
Tools (21 total)
Dispatch Platform — Tier 0 (no auth required)
| Tool | Endpoint | Description |
|---|---|---|
dispatch_health_check |
/healthz |
Service health + snapshot ages |
dispatch_get_feeds |
/api/v1/feeds |
Feed freshness and error state |
dispatch_get_tfr |
/api/v1/tfr |
Active TFRs from FAA |
dispatch_get_tfr_enriched |
/api/v1/tfr-enriched |
TFRs with AI threat interpretation |
dispatch_get_weather |
/api/v1/weather |
DC-area METAR snapshot |
dispatch_get_alerts |
/api/v1/alerts |
Active NWS weather alerts |
dispatch_get_notams |
/api/v1/notams |
Active NOTAMs (requires FAA key on Pi) |
dispatch_get_cps |
/api/v1/cps |
Critical Predictability State (HEMS go/no-go) |
dispatch_get_route |
/api/v1/route |
Ground route impact assessment |
dispatch_get_amtrak |
/api/v1/amtrak |
Amtrak status at WAS |
dispatch_get_brief |
/api/v1/brief |
AI-generated daily operational brief |
dispatch_get_opsplan |
/api/v1/opsplan |
ATCSCC National Operations Plan |
dispatch_get_runsheet |
/api/v1/runsheet |
Active trip runsheet (Tailscale-only) |
Dispatch Platform — Watchlist
| Tool | Method | Description |
|---|---|---|
dispatch_watchlist_get |
GET | List active VIP watchlist sessions |
dispatch_watchlist_add |
POST | Add flight/person to watchlist |
dispatch_watchlist_remove |
DELETE | Remove watchlist session |
Admin (requires DISPATCH_TOKEN)
| Tool | Description |
|---|---|
dispatch_admin_health |
Extended health check |
dispatch_admin_refresh_feed |
Force-refresh a named feed |
dispatch_admin_force_recompute_cps |
Force CPS recomputation |
dispatch_admin_force_opsplan_snapshot |
Force opsplan fetch |
dispatch_admin_send_push_alert |
Send ntfy push notification |
dispatch_admin_get_audit_log |
View audit log |
Flight Tracking — airplanes.live (no auth)
| Tool | Description |
|---|---|
flight_get_by_callsign |
ADS-B position lookup by ICAO callsign |
flight_get_by_registration |
ADS-B position + hex by tail number |
flight_get_by_hex |
ADS-B position by confirmed ICAO hex |
Hex resolution order: callsign → registration → hex. Always confirm hex via
flight_get_by_registration before adding to watchlist — hex is airframe-bound;
callsign-to-hex mappings can be stale day-over-day.
Install
pip install -e .
Requires Python 3.11+.
Configure
| Env var | Default | Notes |
|---|---|---|
DISPATCH_BASE_URL |
https://ops.csexecutiveservices.com |
Use Tailscale IP for runsheet |
DISPATCH_TOKEN |
(empty) | Required for admin tools (csex-token create) |
DISPATCH_TIMEOUT |
30 |
HTTP timeout (seconds) |
ADSB_TIMEOUT |
15 |
airplanes.live timeout (seconds) |
DISPATCH_MCP_TRANSPORT |
stdio |
stdio or http |
DISPATCH_MCP_PORT |
8080 |
Port when transport=http |
Context efficiency and plan compatibility
MCP tool responses are structured and compact — each tool returns only the data the agent actually needs, rather than dumping raw API payloads into the context window. This matters for subscription plan users:
- Claude Pro ($20/mo) — operational dispatch workflows (TFR checks, CPS queries, flight
lookups, daily brief) stay well within the 5-hour message window because context stays lean.
No API key required. Pairing with a context guardian skill (e.g. the
dispatch-context-guardianCowork skill bundled with this deployment) automatically compacts sessions before they hit plan limits, extending sessions significantly without manual intervention. - Claude Max / API — no additional benefit from a guardian on that axis, but the compact responses still reduce per-request token cost and latency.
- Cline / Cursor / Windsurf — same economy applies; MCP tool calls consume far fewer tokens than equivalent REST-then-paste workflows.
The guardian angle is specifically useful for operators running continuous dispatch monitoring inside a chat session — without it, accumulated API responses and skill output would exhaust a Pro plan window in under an hour. With compact tool responses + guardian-triggered compaction, the same session can run for a full shift.
Use with Claude Code
Register at user scope (persists across projects):
claude mcp add dispatch \
-e DISPATCH_BASE_URL=https://ops.csexecutiveservices.com \
-e DISPATCH_TOKEN=your-token-here \
--scope user \
-- dispatch-mcp
Or add to ~/.claude/.claude.json directly under "mcpServers":
{
"mcpServers": {
"dispatch": {
"type": "stdio",
"command": "/full/path/to/dispatch-mcp",
"args": [],
"env": {
"DISPATCH_BASE_URL": "https://ops.csexecutiveservices.com",
"DISPATCH_TOKEN": "your-token-here"
}
}
}
}
Use with Cline / Cursor / Zed / Windsurf
Same MCP JSON config block — each supports mcpServers in their settings file.
Path the command to the installed dispatch-mcp binary or python -m dispatch_mcp.server.
Use with Open WebUI / Conduit
See docs/mcpo-openwebui.md for the mcpo bridge setup.
Verify syntax
python -m py_compile dispatch_mcp/server.py \
dispatch_mcp/config.py \
dispatch_mcp/client.py \
dispatch_mcp/tools/dispatch.py \
dispatch_mcp/tools/flight.py \
dispatch_mcp/tools/admin.py
echo "All clean"
Notes
- Tier 0 endpoints (
/api/v1/*) require no authentication. dispatch_get_runsheetis Tier 1 (Tailscale-gated): setDISPATCH_BASE_URL=http://100.94.80.100:8000.- Admin endpoints require
DISPATCH_TOKEN. Create tokens on the Pi:csex-token create. dispatch.csexecutiveservices.comhas Cloudflare Access on POST routes; useops.csexecutiveservices.comfor programmatic access.- CUI rules: this server never generates or exposes SHARES/HEARS/HEART frequencies. The platform ships with empty placeholder credential files.
License
MIT
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。