Connections

Send an email from the member's own domain

connections_email_send

Destructive

Use this when the member asks to send or email someone from their own business domain (Connections Send). Sends ONE real message, now or at a scheduled time; it lands in the recipient's inbox and cannot be recalled, and dry_run:true returns the exact message without sending it. The sender must be on a domain this workspace has connected and verified. When nothing goes out, the reply is an error naming why: a suppressed recipient, an unverified domain, a spent daily ceiling, or no credits. To check on a sent message use connections_email_get; to see what went out use connections_emails_list.

Input Schema

TableJSON Schema

Name	Required	Description
`cc`	No	Optional carbon-copy recipients.
`to`	Yes	One recipient address, or several.
`bcc`	No	Optional blind-copy recipients.
`from`	Yes	The sender, on a domain this workspace has connected and verified. May be 'Name <you@yours.com>'.
`html`	No	The HTML body. Not with `template`.
`text`	No	The plain-text body. Send text, html, or both. Not with `template`.
`dry_run`	No	true previews the exact call with nothing sent.
`subject`	No	The subject line. Required unless `template` names a stored template that carries one.
`audience`	No	Does this person already know the sender? warm = they know you or asked for this mail; cold = no prior relationship. Required for every recipient, no default: a send with no audience returns an error. One value covers the whole message; recipient_audience sets it per address and wins.
`reply_to`	No	Where replies should go, when that is not the sender.
`template`	No	The name of a template stored in this workspace (e.g. 'order-receipt'), used INSTEAD of text/html. Its placeholder values go in `variables`.
`variables`	No	Values for the template's {{placeholders}}, as a flat object. Every placeholder needs one.
`references`	No	The thread's earlier Message-IDs, oldest first (the References header of the message being answered). Only read with in_reply_to, which is appended when missing.
`in_reply_to`	No	When this answers an email someone sent (a vendor's support ticket, say): that message's Message-ID, e.g. '<abc@zendesk.com>'. The reply then joins their existing thread; without it, their system opens a new one.
`scheduled_at`	No	Optional send time: an ISO instant WITH a timezone ('2026-10-01T09:00:00Z') or a relative time ('in 2 hours'). A time with no timezone returns an error asking for the zone. Omit to send now.
`idempotency_key`	No	Your own key for this message. Sending the same key twice returns the first result instead of sending again - set it whenever a retry is possible.
`recipient_audience`	No	The audience per address, as a flat object: { 'person@yours.com': 'warm' }. Wins over audience for the addresses it names.

Schema Changelog

Changes observed during successful MCP inspections.

Changed1 schema field changedOct 10, 2026

changedInput schema / properties / audience / description

Previous value: -"Does this person already know the sender? warm = they know you or asked for this mail; cold = no prior relationship. Required for every recipient, no default: a send with no audience is refused. One value covers the whole message; recipient_audience sets it per address and wins."New value: +"Does this person already know the sender? warm = they know you or asked for this mail; cold = no prior relationship. Required for every recipient, no default: a send with no audience returns an error. One value covers the whole message; recipient_audience sets it per address and wins."

Changed3 schema fields changedOct 9, 2026

addedInput schema / anyOf

Added value: +[
+  {
+    "required": [
+      "audience"
+    ]
+  },
+  {
+    "required": [
+      "recipient_audience"
+    ]
+  }
+]

addedInput schema / properties / audience

Added value: +{
+  "description": "Does this person already know the sender? warm = they know you or asked for this mail; cold = no prior relationship. Required for every recipient, no default: a send with no audience is refused. One value covers the whole message; recipient_audience sets it per address and wins.",
+  "enum": [
+    "cold",
+    "warm"
+  ],
+  "type": "string"
+}

addedInput schema / properties / recipient_audience

Added value: +{
+  "additionalProperties": {
+    "enum": [
+      "cold",
+      "warm"
+    ],
+    "type": "string"
+  },
+  "description": "The audience per address, as a flat object: { 'person@yours.com': 'warm' }. Wins over audience for the addresses it names.",
+  "type": "object"
+}

Changed2 schema fields changedSep 29, 2026

addedInput schema / properties / in_reply_to

Added value: +{
+  "description": "When this answers an email someone sent (a vendor's support ticket, say): that message's Message-ID, e.g. '<abc@zendesk.com>'. The reply then joins their existing thread; without it, their system opens a new one.",
+  "type": "string"
+}

addedInput schema / properties / references

Added value: +{
+  "description": "The thread's earlier Message-IDs, oldest first (the References header of the message being answered). Only read with in_reply_to, which is appended when missing.",
+  "oneOf": [
+    {
+      "type": "string"
+    },
+    {
+      "items": {
+        "type": "string"
+      },
+      "type": "array"
+    }
+  ]
+}

AddedSep 25, 2026

TDQS

A4.8/5.0

Behavior5/5

Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?

Annotations already declare destructive=true and openWorld=true, but the description goes well beyond them: messages 'cannot be recalled', they land in the recipient's inbox, the sender must be on a connected+verified domain, and failures return a named reason (suppressed recipient, unverified domain, spent daily ceiling, no credits). That is exactly the operational context an agent needs before firing an irreversible send.

Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.

Conciseness5/5

Is the description appropriately sized, front-loaded, and free of redundancy?

Three tightly packed sentences: trigger first, then irreversibility plus the dry-run escape hatch, then failure modes, then sibling routing. Every clause carries information and nothing is repeated from the schema or annotations.

Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.

Completeness4/5

Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?

For a 17-parameter, nested-object, destructive tool with no output schema, the description covers the risky and ambiguous parts (irreversibility, verification prerequisite, error taxonomy, scheduling semantics are in the schema). The one gap is what a successful call returns — no output schema exists, so a phrase about the success payload (message id / status / scheduled confirmation) would close the loop.

Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.

Parameters4/5

Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?

Schema description coverage is 100%, so the baseline is 3; the description still adds meaning by explaining that dry_run:true returns the exact message without sending and by tying the `from` parameter to the verified-domain requirement. The idempotency_key note ('returns the first result instead of sending again') is consistent with a default idempotentHint of false rather than contradicting it, since idempotency only holds when the key is supplied.

Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.

Purpose5/5

Does the description clearly state what the tool does and how it differs from similar tools?

Specific verb+resource ('send or email someone from their own business domain') with the product name (Connections Send) and scope ('ONE real message, now or at a scheduled time'). It clearly separates itself from the read-side siblings connections_email_get and connections_emails_list, which it names explicitly.

Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.

Usage Guidelines5/5

Does the description explain when to use this tool, when not to, or what alternatives exist?

The trigger condition is stated up front ('when the member asks to send or email someone'), and the routing to alternatives for other intents is explicit: connections_email_get to check status, connections_emails_list to see what went out. It also names dry_run:true as the non-sending preview path, so the agent knows when not to actually send.

Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.

All Connections tools Inspect

Try in Browser

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Glama MCP Gateway

Add one secure layer between your agents and this server.

Resources

Need Help?