MCP · stdio · open spec
AI agents that run
your WhatsApp inbox.
Every agent below speaks MCP. Drop one block in its config. Restart it. Now it reads Instagram DMs, drafts WhatsApp replies, and tags hot leads for you.
- Claude CodeThe Claude Code MCP server that connects your terminal to your inbox.Open
- CursorThe Cursor MCP server that connects your editor to your inbox.Open
- Claude DesktopThe Claude Desktop MCP server that connects your inbox to the app.Open
- WindsurfThe Windsurf MCP server that gives Cascade access to your inbox.Open
- ChatGPTBuild a Custom GPT that runs your WhatsApp and Instagram inbox.Open
- ZedThe Zed MCP context server that connects your editor to your inbox.Open
How it works
- 01
Pick an agent below
Choose the AI tool you already use — Claude, Cursor, Windsurf, Zed, or ChatGPT. Each page has a ready-made config block.
- 02
Drop one MCP config block
Paste the snippet into your agent's config file. Add your API key. Restart the app. Setup takes under two minutes.
- 03
Your agent reads DMs, drafts replies, tags leads
From that moment on, your agent can pull live conversations, write contextual replies, and flag high-intent contacts — all from a single prompt.
What your agent can do
- Read every Instagram DM and WhatsApp message
- Draft contextual AI replies using your business knowledge
- Tag high-intent leads automatically
- Escalate sensitive conversations to a human
- Pull conversation history as context
Why this exists
Most AI agents are isolated from your customers. InstantReply's MCP server bridges that gap — your agent reads live WhatsApp and Instagram conversations, drafts replies with your business context loaded, and hands back to your team when human judgment matters. Works with any MCP-compatible client. Free on every InstantReply plan.
Don't see your tool?
Any MCP-compatible client works.
The server speaks plain MCP over stdio. If your tool can run an npx command, it can run InstantReply.