Resend Agent for OpenBot

This is an OpenBot agent-kind plugin that provides email sending capabilities using Resend and AI.

Features

• Natural language interface for sending emails.
• Support for CC, BCC, and Reply-To.
• Optional attachments (base64 / data URL, or a public HTTPS URL per file).
• Retrieve status of sent emails.
• Powered by OpenAI (GPT-4o).

Configuration

You need both a Resend API key (to send mail) and an OpenAI API key (for the LLM). You can set them in AGENT.md, or store them as OpenBot variables (RESEND_API_KEY, OPENAI_API_KEY), or mix both (e.g. put OpenAI in config and only add Resend via the setup form).

If either key is missing when the agent runs, a form widget asks only for what is still needed—so if OpenAI is already configured, you are only prompted for Resend.

plugins:
  - name: "@meetopenbot/resend"
    config:
      resendApiKey: "re_123456789"
      openaiApiKey: "sk-proj-..."   # omit if provided elsewhere
      defaultFrom: "Your Name <[email protected]>"

Sending emails with attachments

Ask the agent to send an email and attach files. The model calls Resend with one of these per attachment:

1. Inline bytes — Paste or provide base64 (raw or a data:mime/type;base64,... data URL). Good for small files the model can embed in the tool call.
2. URL — A public HTTPS URL (path) Resend can fetch. Better for larger files or anything already hosted (S3, CDN, signed link, etc.).

Each attachment needs a filename (e.g. report.pdf). Optionally set contentType (MIME type) and contentId for inline HTML images (<img src="cid:yourId">).

Resend enforces per-email size limits (on the order of tens of MB total); prefer URLs for big binaries.

Example prompts

• “Email [email protected] with subject ‘Q4 summary’ and attach https://example.com/public/deck.pdf as deck.pdf.”
• “Send the weekly update to me; attach this CSV [paste base64 or describe a public link].”

How it works

This is an agent-kind plugin, meaning it handles the conversation directly. When you talk to the agent, it uses an LLM to understand your request and calls the Resend API internally to perform actions like sending emails or checking email status.

Development

1. Install dependencies:

   npm install

2. Build the plugin:

   npm run build

License

MIT