AI agents use send_draft to create or update resources in Gmail MCP — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Gmail MCP environment.
This tool creates irreversible communication by transmitting an email to one or more recipients. While not destructive in the sense of deleting data, sending an email is a Write action that creates new data (sent message records) and effects that cannot be easily undone—emails cannot be reliably 'unsent' once delivered.
From the tool's definition Tool name 'send_draft' and description 'Send an existing draft' indicate the tool sends an email message. Sending emails modifies the user's sent folder and the recipient's inbox, creating permanent communication records.
Documented attack patterns abuse exactly the kind of access send_draft gives an agent:
PolicyLayer is an MCP gateway — it sits between your AI agents and Gmail MCP, and nothing reaches the server without passing your rules. This is the rule we recommend for send_draft:
{
"version": "1",
"default": "deny",
"tools": {
"send_draft": {
"limits": [
{
"counter": "send_draft_rate",
"window": "minute",
"max": 30,
"scope": "grant"
}
]
}
}
}send_draft stays usable, but capped — an agent stuck in a loop can't make hundreds of changes a minute. Everything else on the server is denied unless you say otherwise.
Free to start. No card required.
Send an existing draft. It is categorised as a Write tool in the Gmail MCP MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.
Register the Gmail MCP server in PolicyLayer and add a rule for send_draft: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Gmail MCP. Nothing to install.
send_draft is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.
Yes. Add a rate_limit block to the send_draft rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.
Set action: deny in the PolicyLayer policy for send_draft. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.
send_draft is provided by the Gmail MCP server (shinzo-labs/gmail-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.
Deterministic rules across all 65 Gmail MCP tools. Per-identity grants. Full audit log. Live in minutes. Nothing to install.
Free to start. No card required.
65 Gmail MCP tools catalogued and risk-classified — across an index of 42,500+ MCP servers.