For AI agents
Let agents run your travel ops.
One CLI. ~49 tools. Your agent searches inventory, places holds, tickets bookings, settles on-chain in USDC, audits every step. Anything you do in Sendero, it can do too.
~ $ npx @sendero/cli@latest
Sendero
v0.1.0 · agent@workspace · Sendero Travel Ops
~ $
Install
Pick your install path. Same auth gate, same tool surface — different ergonomics.
The fastest path. One npx command, no global install. Auth runs in your browser and writes ~/.sendero/key with chmod 600.
1 · Mint and save an API keybash
npx @sendero/cli@latest auth login
2 · Bootstrap the Claude Code pluginbash
npx @sendero/cli@latest mcp install
3 · List the live tool catalogbash
npx @sendero/cli@latest tools list
4 · Dispatch a tool (JSON-RPC over /api/mcp)bash
npx @sendero/cli@latest tools call search_flights '{"origin":"BUE","destination":"MIA","date":"2026-05-12"}'Endpoint override (advanced)
Set SENDERO_API_URL to a non-prod base (e.g., staging). The CLI reuses it for both /api/openapi.json and /api/mcp. Set SENDERO_API_KEY=ak_your_key_here to bypass the saved file (useful in CI).
CLI
- ◇ Search, hold, confirm, settle, refund
- ◇ Structured JSON when piped, tables on TTY
- ◇ Browser-based key mint
- ◇ Workspace switching
- ◇ Single npx command
MCP
- ◇ ~49 tools across travel-ops
- ◇ HTTP transport — works with any MCP client
- ◇ Claude Desktop, Claude Code, Cursor, Codex, VS Code
- ◇ Same auth gate as the CLI
- ◇ Sandbox keys auto-mint on workspace create
Developer experience
- ◇ OpenAPI 3.1 spec at /api/openapi.json
- ◇ llms.txt advertises every surface
- ◇ TypeScript SDK auto-generated
- ◇ One canonical tool registry
- ◇ Open-source plugin + CLI
Get started
One CLI. One MCP server. Every travel-ops operation your agent needs.