AI sales agents for founders: suggest, edit and read outreach agents. A person approves every send.
Maintain this server? Claim it
Call this server's tools from your browser. Nothing to install.
A project token from the Moab app (Connect, Coding agent), sent as a Bearer token.
Requests go through mcptry because browsers can't call most MCP servers directly. Header values are sent to the server for that request only and are never stored or logged.
Tool list from our last check. Connect live to refresh it.
Edit the copy (subject, body, name) or the sequence (steps: visit, invite with a note of 200 characters, the message, follow-ups; notes and follow-ups may use only {{first_name}} and {{company}}) of a sandbox or dealt agent. A live agent is 409 agent_live: only a person changes it, in the app. Never the sender, the cap or the status.
This tool isn't marked read-only, so it may change something on the server or in an account you connected.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http moab https://moabcrm.com/mcp \ --header 'Authorization: Bearer YOUR_TOKEN'
Headers
Authorization (required): A project token from the Moab app (Connect, Coding agent), sent as a Bearer token.Edit the copy (subject, body, name) or the sequence (steps: visit, invite with a note of 200 characters, the message, follow-ups; notes and follow-ups may use only {{first_name}} and {{company}}) of a sandbox or dealt agent. A live agent is 409 agent_live: only a person changes it, in the app. Never the sender, the cap or the status.
| Argument | Type | Description |
|---|---|---|
| body | string | |
| name | string | |
| steps | array | |
| agentId* | string | |
| subject | string |
{
"type": "object",
"required": [
"agentId"
],
"properties": {
"body": {
"type": "string"
},
"name": {
"type": "string"
},
"steps": {
"type": "array",
"items": {
"type": "object",
"required": [
"kind",
"waitDays",
"body"
],
"properties": {
"body": {
"type": "string",
"description": "invite: the note ('' = none); follow_up: its text; visit and message: ''"
},
"kind": {
"enum": [
"visit",
"invite",
"message",
"follow_up"
]
},
"waitDays": {
"type": "integer",
"description": "days after the step before (after an invite: after they accept)"
}
},
"additionalProperties": false
}
},
"agentId": {
"type": "string"
},
"subject": {
"type": "string"
}
},
"additionalProperties": false
}Deal more agent cards for this project (a background job; call again later and read the deck with moab_experiments_context). size 1 to 5.
| Argument | Type | Description |
|---|---|---|
| like | string | an agent id to deal more like |
| size | integer | |
| steer | any |
{
"type": "object",
"required": [],
"properties": {
"like": {
"type": "string",
"description": "an agent id to deal more like"
},
"size": {
"type": "integer"
},
"steer": {
"enum": [
"more_like_this",
"different_buyer",
"different_message"
]
}
},
"additionalProperties": false
}Swipe a dealt card: right puts it in the sandbox (drafts go only to the founder), left passes it. Never sends anything.
| Argument | Type | Description |
|---|---|---|
| agentId* | string | |
| decision* | any |
{
"type": "object",
"required": [
"agentId",
"decision"
],
"properties": {
"agentId": {
"type": "string"
},
"decision": {
"enum": [
"right",
"left"
]
}
},
"additionalProperties": false
}What is being tested and what the last 7 days counted: agents, their numbers and verdicts, what is working.
No arguments.
{
"type": "object",
"required": [],
"properties": {},
"additionalProperties": false
}Recent replies from prospects (text only, newest first).
| Argument | Type | Description |
|---|---|---|
| limit | integer |
{
"type": "object",
"required": [],
"properties": {
"limit": {
"type": "integer"
}
},
"additionalProperties": false
}Which senders are connected and healthy. No secrets.
No arguments.
{
"type": "object",
"required": [],
"properties": {},
"additionalProperties": false
}The message templates agents use, with the evidence behind each.
No arguments.
{
"type": "object",
"required": [],
"properties": {},
"additionalProperties": false
}The project this token is for.
No arguments.
{
"type": "object",
"required": [],
"properties": {},
"additionalProperties": false
}We read every tool's name, description and inputs the way the model will, looking for instructions hidden from you. A clean result means no warning signs, not a guarantee. Scanned Nov 20, 2026.
No readOnlyHint / destructiveHint annotations, so clients can't tell a read from a write without asking.
Fix: Add annotations to each tool, e.g. { readOnlyHint: true } for tools that only read.