// ai agents and mcp
Via APIZed email verification with pingutil
Zed's agent can write the email handling in your app, but it has no source of truth for whether an address is live or disposable. Adding pingutil as a remote context server in Zed's settings gives the agent the three pingutil tools, configured with a URL and an Authorization header.
Where it connects: Zed context_servers remote MCP server (Settings > AI > MCP Servers > Add Remote Server, url plus headers)
What goes wrong without it
- The agent in Zed wrote an "is disposable" helper that checks five domains.
- I want to see what the verification API returns for a few addresses without leaving the editor.
- Our test fixtures use made-up addresses and nobody knows which ones are real people.
What you can do with Zed and pingutil
Write fixtures from real responses
Ask the agent to call lookup_email and verify_email on a handful of sample addresses and save the results as fixtures, so your tests match the real fields (status, reason, category, verdict.is_spam).
Audit addresses in test data
Have the agent verify the addresses in your fixtures and swap any that belong to real, deliverable mailboxes for example.com addresses.
Classify domains in a config change
When you edit an allow list or a list of partner domains, ask the agent to classify each one and flag free webmail or disposable entries.
Set it up
- 1
Create an API key in the pingutil dashboard.
- 2
In Zed open Settings > AI > MCP Servers, click Add Server, then Add Remote Server.
- 3
Enter the URL https://pingutil.com/mcp and an Authorization header with the value Bearer followed by your key; Zed writes the entry to context_servers in your settings file.
- 4
Keep the entry in your user settings, not a project .zed/settings.json, because Zed's docs show no environment variable syntax for headers and the key is stored as typed.
- 5
Open the Agent Panel and ask the Zed agent to verify or classify addresses with pingutil.
{
"context_servers": {
"pingutil": {
"url": "https://pingutil.com/mcp",
"headers": {
"Authorization": "Bearer pk_live_your_key_here"
}
}
}
}What to do with each result
| pingutil result | Action in Zed |
|---|---|
| verify_email: status valid on a fixture address | Replace it with an example.com address so tests never email a real person. |
| status invalid | Safe to use as the "bad address" case in tests. |
| lookup_email: verdict.is_spam true | Use as the disposable or randomized case in the signup test. |
| lookup_email: category free or disposable in an allow list | Flag the entry for review in the change. |
Limits to plan for
- Zed's docs show the header value as a literal and document no environment variable interpolation for headers, so the key sits in plain text in your settings file. Keep that file out of shared dotfile repos.
- MCP servers configured in Zed are used by the Zed agent; external agents run from Zed (such as a CLI in a terminal thread) read their own MCP configuration.
- Slow mail servers can make a mailbox check take up to about 25 s, and calls count toward the key's Free plan limits (300 verifications a day and 3,000 a month, 2,000 lookups a day and 30,000 a month). No OAuth.
Zed documentation
Zed is a trademark of its owner. This integration uses Zed's documented extension points; it is not an official Zed app.
Common questions
- Does the Zed agent see my API key?
- Zed adds the header to its requests to pingutil; the agent works with tool results. Because the key is stored in your settings file, avoid asking the agent to read or edit that file.
- Do I need a local process or bridge?
- No. Zed connects to remote MCP servers by URL, and pingutil is a hosted Streamable HTTP server.
- Which tools does the server expose?
- verify_email, verify_emails (up to 50 addresses) and lookup_email, all read-only.
Stop bad addresses in Zed today
Sign in with your email, create an API key and paste it into the step above. Test on a few real addresses first: the live demo shows exactly what your Zed setup will receive.