Pidgeon for JavaScript and TypeScript
Email inboxes for AI agents. Pidgeon is the email inbox API for AI agents. It gives agents their own inboxes so they can work with emails just like a human.
npm install @pidgeon-ai/sdk
接入你的智能体
發布者自己的指令。不是 PlugMCP 端點。
用我們的金鑰與日誌託管一份副本。
npx -y @pidgeon-ai/[email protected]Claude Code 沒有本機指令的安裝連結。複製下方指令並在終端機中執行。
claude mcp add pidgeon-ai-cli -- npx -y @pidgeon-ai/cli@1.44.0在終端機中執行。
{ "mcpServers": { "pidgeon-ai-cli": { "command": "npx", "args": [ "-y", "@pidgeon-ai/[email protected]" ] } }}claude_desktop_config.json
遠端 URL 還是本機指令? 兩種安裝方式的差異
來自發布者
Email inboxes for AI agents. Pidgeon is the email inbox API for AI agents. It gives agents their own inboxes so they can work with emails just like a human.
npm install @pidgeon-ai/sdk
Email inboxes for AI agents, from a terminal — the Pidgeon CLI: create addresses, send, receive and reply.
Create an address, send from it, wait for the answer, and reply in the thread:
import {
client,
identitiesCreate,
messagesAwait,
messagesReply,
messagesSend,
} from '@pidgeon-ai/sdk';
client.setConfig({ auth: () => process.env.PIDGEON_API_KEY! });
// 1. An address of its own. Any free name on pidgeon.email works with no DNS.
const { data: address } = await identitiesCreate({
body: { address: '[email protected]', type: 'agent' },
throwOnError: true,
});
// 2. Send from it.
await messagesSend({
body: {
from: address.data.email,
to: '[email protected]',
subject: 'Staging access',
text: 'Could you send me the access code?',
},
throwOnError: true,
});
// 3. Receive: one held-open request until the answer arrives, up to 60 seconds.
const { data: awaited } = await messagesAwait({
query: { identity_id: address.data.id, timeout: 60 },
throwOnError: true,
});
// 4. Reply into the same conversation. Recipients and subject come off the original.
if (awaited.data.message) {
console.log(awaited.data.message.extracted.codes);
await messagesReply({
path: { id: awaited.data.message.id },
body: { text: 'Thanks, that worked.' },
throwOnError: true,
});
}
PIDGEON_API_KEY is a pgn_live_… key. Make one at
pidgeon.ai/app/developers/keys, or let
your agent sign itself up: pidgeon.ai/skill.md.
examples/openai-agents is an OpenAI Agents SDK
agent on the hosted server, examples/claude-agent-sdk
a Claude Agent SDK one, and examples/mastra a Mastra one. None
needs a package of ours. examples/ai-sdk is a Vercel AI SDK
agent on @pidgeon-ai/ai-sdk, asking before it replies.
activepieces/pidgeon is the Activepieces piece,
@pidgeon-ai/piece-pidgeon, in the layout Activepieces' own repository
uses (setup).
An MCP client needs no package at all. Point it at
https://pidgeon.ai/api/mcp and it signs you in with OAuth:
claude mcp add --transport http pidgeon https://pidgeon.ai/api/mcp
The plugin for Claude Code, Codex, Cursor and Gemini CLI is pidgeon-ai/pidgeon-plugins.
The SDK and the CLI are generated from openapi.json, which
is the contract the API validates against and is published at
https://pidgeon.ai/openapi.json. The application that serves it is closed
source; these clients are not, and this is where they are readable. This
repository is copied from that one on every release.
So a pull request against generated code cannot be merged as written — the fix belongs in the schema it came from. Open an issue instead and say what is wrong; that is the faster route either way, because a wrong field here is a wrong field in the published contract and in the Python client too.
Issues and questions: open one, or email [email protected].
MIT licensed.
Email inboxes for AI agents, from a terminal — the Pidgeon CLI: create addresses, send, receive and reply.
{ "mcpServers": { "pidgeon-ai-cli": { "command": "npx", "args": [ "-y", "@pidgeon-ai/[email protected]" ] } }}工具
尚無工具清單。遠端伺服器會在首次健康檢查後取得工具清單。
常見問題
在本頁複製適用於 Cursor、Claude Desktop、Claude Code 或 VS Code 的 stdio 設定。它使用發布者的安裝方式,而不是 PlugMCP 代理伺服器。
這筆記錄未列出必要的環境變數。
@pidgeon Ai/cli 透過 stdio 在本機執行。
上次檢查時間:2026-11-14。 Email inboxes for AI agents, from a terminal — the Pidgeon CLI: create addresses, send, receive and reply.