Mailgent for Cohere
Use Mailgent with Cohere's tool use. Your Command-powered agent gets a real inbox and identity to send mail, hold secrets, and act on what it decides.
Cohere's Command models support tool use, so Mailgent connects through the same mechanism as other frameworks. Define the actions as tools and the model calls them when its reasoning requires.
The result is a Cohere-powered agent with a real address and the surrounding identity — able to email, store secrets, and operate accountably.
Define tools for Command
Provide tool specifications for the Mailgent actions you allow. When the model returns a tool call, run it against the API with the agent's key and pass the result back into the conversation.
Scopes on the key keep the agent within bounds regardless of what the model attempts.
tools = [{
"name": "mail_send",
"description": "Send email from the agent's inbox",
"parameter_definitions": {
"to": {"type": "str", "required": True},
"subject": {"type": "str", "required": True},
"text": {"type": "str", "required": True} }
}]Accountable actions
Each tool call routes through the agent's scoped identity, so the model's reach is exactly what you grant — and every action is logged.
FAQ
Which Cohere models support this?
Command models with tool use can drive Mailgent tools.
Can I restrict the agent to read-only?
Yes. Grant only read scopes and the agent can list and read mail but not send.
Give your agent an inbox.
A real email address, a vault, 2FA, and an identity in one API call.