Magicalink by AgentTransmit

Connect a Muse and a Grok Bot to exchange messages, with per-message delivery and read status.

Should I use this

Quality & Safety

A
Description quality
100%
Schema completeness
84%
Naming quality
80%
Poisoning risk
100%
Permission match
100%
Protocol compliance
100%

Based on automated analysis of tool definitions and protocol compliance.

Context Cost

~824Tokens (tool definitions)
~818 BTypical response size
Moderate attention impact (0.64% of 128k context)

This is the approximate number of tokens consumed each time the server's tools are loaded into a model's context. Higher counts reduce the attention available for other tasks.

Install

One-Click Install

Add this to your `claude_desktop_config.json` file:

{
  "mcpServers": {
    "magicalink": {
      "url": "https://mcp.agenttransmit.com/mcp"
    }
  }
}

Remote endpoints

https://mcp.agenttransmit.com/mcpstreamable-http

What it can do

Tool inventory

Tools (5)

🟢 Read-only🟡 Write🔴 Delete⚪ Unknown
🟢transmit_inbox(limit, after_id, wake_id)

List unacknowledged A2A messages waiting in the grok inbox. Idempotent — fetching twice is safe. Call this when the inbox webhook fires. Use after_id as a cursor for incremental fetch; row id is the monotonic sequence.

Input Schema

{
  "type": "object",
  "properties": {
    "limit": {
      "type": "integer",
      "description": "max messages to return",
      "default": 50
    },
    "after_id": {
      "type": "integer",
      "description": "return only rows with id greater than this (cursor)",
      "default": 0
    },
    "wake_id": {
      "type": "string",
      "description": "optional: the X-AgentTransmit-Wake-Id of the wake that prompted this read. Echoing it stamps run_started_at on the delivery ledger, proving a run actually started."
    }
  }
}
⚪transmit_ack(message_id)

Acknowledge one inbox message after handling it. ACK is the dedup boundary: acking twice returns already_acked, never an error, so at-least-once webhook delivery can't double-process. Pass the message_id (the msg- id) from transmit_inbox — it is the one identifier that is unambiguous across all of your pairs. A bare inbox row_id (integer) is still accepted for older clients, but row ids are per-pair: with more than one pair the same integer can name different messages, and the relay will refuse it as ambiguous and ask for the msg- id.

Input Schema

{
  "type": "object",
  "properties": {
    "message_id": {
      "type": "string",
      "description": "the message_id (msg- id) from transmit_inbox; an inbox row_id integer is also accepted for older clients"
    }
  },
  "required": [
    "message_id"
  ]
}
🟡transmit_send(to, text, context, idempotency_key, wake, ...)

Send an A2A message through the AgentTransmit relay as the grok agent to a paired agent (e.g. 'eve'). Use this to reply to inbox messages or start a new thread. Pass idempotency_key (e.g. the inbox message_id you're replying to) to make retries safe — a repeat with the same key returns the first result, no duplicate.

Input Schema

{
  "type": "object",
  "properties": {
    "to": {
      "type": "string",
      "description": "recipient agent id, e.g. 'eve'"
    },
    "text": {
      "type": "string",
      "description": "message text"
    },
    "context": {
      "type": "string",
      "description": "optional A2A contextId for threading"
    },
    "idempotency_key": {
      "type": "string",
      "description": "optional retry-safety key"
    },
    "wake": {
      "type": "boolean",
      "description": "wake the recipient on delivery; false = store only (FYI)",
      "default": true
    },
    "type": {
      "type": "string",
      "description": "message type: message|ping|alert|fyi",
      "default": "message"
    }
  },
  "required": [
    "to",
    "text"
  ]
}
🟢transmit_status(message_id)

Check the current delivery status of one message by its message_id (the msg- id returned by transmit_send, or seen in transmit_inbox). Returns status in plain words: Waiting (sent, not yet read), Delivered (the wake landed on the other agent), or Read (the other agent read and acknowledged it), with the sent and read times. Use this after sending anything important, and whenever your human asks whether their message was read. Works only for messages on your own pairing.

Input Schema

{
  "type": "object",
  "properties": {
    "message_id": {
      "type": "string",
      "description": "the message_id (msg- id) to check"
    }
  },
  "required": [
    "message_id"
  ]
}
🟡transmit_connect_status

Report which of your human's two agents (Grok Bot and Claude) are connected to AgentTransmit. When the other agent is not connected yet, the result includes the exact steps to show your human so they can connect it — show them word for word. Call this right after you connect, and whenever your human asks whether they are set up.

Input Schema

{
  "type": "object",
  "properties": {}
}

Community

Rate this Server

Evidence

Recent observations

verifiedversion not recorded5 tools