paperplane-mcp
MCP connector for Paperplane that lets AI tools page you on your phone with event notifications, blocking questions, alert management, replies, and follow-up ticket filing using a single long-lived token.
README
paperplane-mcp
An MCP connector for Paperplane — the inbox for the events your machines send you.
It lets an AI tool page you on your phone when a long job finishes or needs a decision, read back what fired, acknowledge or resolve an alert, and file a follow-up ticket. One long-lived credential, no per-channel keys.
Setup
1. Create a token
In the Paperplane app: Settings → Developer tokens → Create token. The secret is shown once —
copy it then. It looks like pplt_ followed by 64 hex characters.
2. Point a client at it
Claude Code — .mcp.json in your project, or ~/.claude.json for every project:
{
"mcpServers": {
"paperplane": {
"command": "npx",
"args": ["-y", "paperplane-mcp"],
"env": {
"PAPERPLANE_TOKEN": "pplt_...",
"PAPERPLANE_DEFAULT_CHANNEL": "Builds"
}
}
}
}
Claude Desktop uses the same block inside claude_desktop_config.json.
| Variable | Required | Default |
|---|---|---|
PAPERPLANE_TOKEN |
yes | — |
PAPERPLANE_API_BASE_URL |
no | https://paperplane-backend.parikshithv01.workers.dev |
PAPERPLANE_DEFAULT_CHANNEL |
no | none — tools then need an explicit channel |
The server checks the token at startup, so a revoked one fails fast with a message on stderr rather than surfacing mid-conversation.
Tools
| Tool | What it does |
|---|---|
paperplane_send_event |
Push a notification to the phone. Supports priority, nature, and action buttons. |
paperplane_ask |
Send a question to the phone and block until it's answered (or times out). |
paperplane_check_ask |
Non-blocking read of an ask's current status — recover an answer after a client-side timeout. |
paperplane_list_channels |
List the channels you belong to. |
paperplane_list_messages |
Read recent events and replies, newest first, with alert status. |
paperplane_set_alert_state |
Mark a message open, acked, or resolved. |
paperplane_reply |
Post a reply into a channel thread. |
paperplane_list_tickets |
Read the personal ticket board. |
paperplane_create_ticket |
File a follow-up, optionally linked back to an alert. |
Making something an alert
An event becomes an alert — badged, counted under "Needs attention", requiring acknowledgement —
when nature is one of error, failure, failed, critical, pending, or priority is one of
high, crit, critical, urgent. Anything else is an ordinary message.
Action buttons
callback attaches buttons to the notification:
{
"title": "Deploy failed",
"message": "api-service · main @ abc123",
"nature": "error",
"callback": [{ "label": "Re-run pipeline", "webhook": "https://ci.example.com/hooks/rerun" }]
}
The phone fires the webhook when the button is pressed — the connector never calls it.
Scopes
A token carries scopes; the connector uses events:write, channels:read, messages:read,
messages:write, tickets:read, and tickets:write. Tokens minted from the app carry all of them.
Mint a narrower one via the API if you want a send-only credential:
curl -X POST "$PAPERPLANE_API_BASE_URL/tokens" \
-H "authorization: Bearer $SESSION_JWT" -H 'content-type: application/json' \
-d '{"name":"CI notifier","scopes":["events:write"]}'
Channel administration, membership, retention, and token management are session-only and stay out of reach of any token.
Development
npm install
npm run build
npm run inspect # builds, then opens the MCP Inspector
Point it at a local backend with PAPERPLANE_API_BASE_URL=http://localhost:8787.
Tool handlers in src/tools/ take parsed arguments and return plain values — they know nothing
about MCP framing, so mounting them on a remote HTTP transport later is additive rather than a
rewrite.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.
E2B
Using MCP to run code via e2b.