iMessage for AI agents, from your iPhone. No Mac needed, and you choose which chats they read.
Maintained by@3eif(verified)Also a maintainer?
Call this server's tools from your browser. Nothing to install.
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.
Count messages, words, or attachments grouped by one or two of sender, conversation, day, week, month, hour, or weekday, in the person's time zone, for one conversation or all, filtered by date (after, before), sender, from_me, has_attachments, or has_reactions. It takes no query or pattern. Use for trends and comparisons instead of reading whole conversations: messages per month in one conversation to see if it changed, per sender in a group chat to see who writes most, per hour to see when people text.
One or two of: sender, conversation, day, week, month, hour, weekday. Times use the person's local time, following daylight saving changes. sender groups by display name across conversations, keeping the person apart from anyone who shares their name; add conversation to keep same-named people in different conversations apart.
Only messages sent at or after this time: RFC 3339, or YYYY-MM-DD for the start of that day in the person's time zone.
Only messages sent before this time: RFC 3339, or YYYY-MM-DD for the start of that day in the person's time zone.
messages (default), words, or attachments.
Only messages whose sender's name contains this text, ignoring case.
True for only the person's own messages, false for only everyone else's.
Count one conversation by its ID. Omit to count every shared conversation.
True for only messages someone reacted to (a tapback), false for only messages without reactions.
True for only messages with attachments, false for only messages without.
Pick your client. Values in capitals, like YOUR_TOKEN, are yours to fill in.
Run in a terminal
claude mcp add --transport http imessage https://mcp.misento.app
Count messages, words, or attachments grouped by one or two of sender, conversation, day, week, month, hour, or weekday, in the person's time zone, for one conversation or all, filtered by date (after, before), sender, from_me, has_attachments, or has_reactions. It takes no query or pattern. Use for trends and comparisons instead of reading whole conversations: messages per month in one conversation to see if it changed, per sender in a group chat to see who writes most, per hour to see when people text.
| Argument | Type | Description |
|---|---|---|
| after | string | null | Only messages sent at or after this time: RFC 3339, or YYYY-MM-DD for the start of that day in the person's time zone. |
| before | string | null | Only messages sent before this time: RFC 3339, or YYYY-MM-DD for the start of that day in the person's time zone. |
| metric | any | messages (default), words, or attachments. |
| sender | string | null | Only messages whose sender's name contains this text, ignoring case. |
| from_me | boolean | null | True for only the person's own messages, false for only everyone else's. |
| group_by* | array | One or two of: sender, conversation, day, week, month, hour, weekday. Times use the person's local time, following daylight saving changes. sender groups by display name across conversations, keeping the person apart from anyone who shares their name; add conversation to keep same-named people in different conversations apart. |
| conversation | string | null | Count one conversation by its ID. Omit to count every shared conversation. |
| has_reactions | boolean | null | True for only messages someone reacted to (a tapback), false for only messages without reactions. |
| has_attachments | boolean | null | True for only messages with attachments, false for only messages without. |
{
"type": "object",
"$defs": {
"Metric": {
"oneOf": [
{
"type": "string",
"const": "messages",
"description": "Number of messages. The default."
},
{
"type": "string",
"const": "words",
"description": "Number of words in message text."
},
{
"type": "string",
"const": "attachments",
"description": "Number of attachments."
}
]
},
"Dimension": {
"oneOf": [
{
"type": "string",
"const": "sender",
"description": "The sender's name."
},
{
"type": "string",
"const": "conversation",
"description": "The conversation's name."
},
{
"type": "string",
"const": "day",
"description": "YYYY-MM-DD."
},
{
"type": "string",
"const": "week",
"description": "ISO week, YYYY-Www."
},
{
"type": "string",
"const": "month",
"description": "YYYY-MM."
},
{
"type": "string",
"const": "hour",
"description": "Hour of the day, 00 to 23."
},
{
"type": "string",
"const": "weekday",
"description": "Monday to Sunday."
}
]
}
},
"$schema": "https://json-schema.org/draft/2020-12/schema",
"required": [
"group_by"
],
"properties": {
"after": {
"type": [
"string",
"null"
],
"description": "Only messages sent at or after this time: RFC 3339, or YYYY-MM-DD for the\nstart of that day in the person's time zone."
},
"before": {
"type": [
"string",
"null"
],
"description": "Only messages sent before this time: RFC 3339, or YYYY-MM-DD for the start\nof that day in the person's time zone."
},
"metric": {
"anyOf": [
{
"$ref": "#/$defs/Metric"
},
{
"type": "null"
}
],
"description": "messages (default), words, or attachments."
},
"sender": {
"type": [
"string",
"null"
],
"description": "Only messages whose sender's name contains this text, ignoring case."
},
"from_me": {
"type": [
"boolean",
"null"
],
"description": "True for only the person's own messages, false for only everyone else's."
},
"group_by": {
"type": "array",
"items": {
"$ref": "#/$defs/Dimension"
},
"description": "One or two of: sender, conversation, day, week, month, hour, weekday.\nTimes use the person's local time, following daylight saving changes.\nsender groups by display name across conversations, keeping the person\napart from anyone who shares their name; add conversation to keep\nsame-named people in different conversations apart."
},
"conversation": {
"type": [
"string",
"null"
],
"description": "Count one conversation by its ID. Omit to count every shared conversation."
},
"has_reactions": {
"type": [
"boolean",
"null"
],
"description": "True for only messages someone reacted to (a tapback), false for only\nmessages without reactions."
},
"has_attachments": {
"type": [
"boolean",
"null"
],
"description": "True for only messages with attachments, false for only messages without."
}
}
}Find conversations and see them at a glance; it gives the conversation IDs the other tools take. To find someone, pass query with part of a name, phone number, or email: {"query": "dad"} returns only matching chats. To read someone's messages, call misento_messages with conversation_query instead, which finds and reads in one call. Each conversation shows when the person last wrote and when others last wrote, total messages, and counts for the last 7 days, the last 30 days, and the 30 days before that. Filter by is_group, last_sender (others: someone else wrote last), or silent_for_days; sort by recent, oldest (longest silence first), or most_active. response_format detailed adds each conversation's last three messages, to judge whether a reply is needed. 20 per page; follow next_cursor for more.
| Argument | Type | Description |
|---|---|---|
| sort | any | recent (default), oldest, or most_active. |
| limit | integer | null | Conversations per page, 1 to 500. Defaults to 20. |
| query | string | null | Only conversations whose name, or a participant's name, has a word that starts with this text ("dad" finds Dad, not Haddad), or whose participant's phone number or email contains it, ignoring case. Phone numbers also match by digits, so 555-0134 finds +15550134. |
| cursor | string | null | The next_cursor from the previous page, with the same other filters and sort. |
| is_group | boolean | null | True for only group chats, false for only one-on-one conversations. |
| last_sender | any | Only conversations whose last message is from `me` or from `others`. |
| response_format | any | concise (default): counts and last-message times. detailed: also each conversation's first message time, when its copy last changed, and its last three messages, long ones cut short. |
| silent_for_days | integer | null | Only conversations where no one has written for at least this many days. |
{
"type": "object",
"$defs": {
"LastSender": {
"oneOf": [
{
"type": "string",
"const": "me",
"description": "The person."
},
{
"type": "string",
"const": "others",
"description": "Anyone else."
}
],
"description": "Who sent a conversation's last message."
},
"OverviewSort": {
"oneOf": [
{
"type": "string",
"const": "recent",
"description": "Most recent message first. The default."
},
{
"type": "string",
"const": "oldest",
"description": "Longest since the last message first."
},
{
"type": "string",
"const": "most_active",
"description": "Most messages in the last 30 days first."
}
]
},
"ResponseFormat": {
"oneOf": [
{
"enum": [
"detailed"
],
"type": "string"
},
{
"type": "string",
"const": "concise",
"description": "The default."
}
],
"description": "How much a response carries; each tool says what that means for it."
}
},
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"sort": {
"anyOf": [
{
"$ref": "#/$defs/OverviewSort"
},
{
"type": "null"
}
],
"description": "recent (default), oldest, or most_active."
},
"limit": {
"type": [
"integer",
"null"
],
"format": "uint32",
"minimum": 0,
"description": "Conversations per page, 1 to 500. Defaults to 20."
},
"query": {
"type": [
"string",
"null"
],
"description": "Only conversations whose name, or a participant's name, has a word that\nstarts with this text (\"dad\" finds Dad, not Haddad), or whose\nparticipant's phone number or email contains it, ignoring case. Phone\nnumbers also match by digits, so 555-0134 finds +15550134."
},
"cursor": {
"type": [
"string",
"null"
],
"description": "The next_cursor from the previous page, with the same other filters\nand sort."
},
"is_group": {
"type": [
"boolean",
"null"
],
"description": "True for only group chats, false for only one-on-one conversations."
},
"last_sender": {
"anyOf": [
{
"$ref": "#/$defs/LastSender"
},
{
"type": "null"
}
],
"description": "Only conversations whose last message is from `me` or from `others`."
},
"response_format": {
"anyOf": [
{
"$ref": "#/$defs/ResponseFormat"
},
{
"type": "null"
}
],
"description": "concise (default): counts and last-message times. detailed: also each\nconversation's first message time, when its copy last changed, and its\nlast three messages, long ones cut short."
},
"silent_for_days": {
"type": [
"integer",
"null"
],
"format": "uint32",
"minimum": 0,
"description": "Only conversations where no one has written for at least this many\ndays."
}
}
}Pull messages, as one timeline or grouped per conversation, newest or oldest first, paged with next_cursor. Combine freely: conversations by ID, or conversation_query to find them by name in the same call (omit both for all), a date range with after and before, a keyword query (match all words, or any of them to try several synonyms at once), a regex pattern, sender, from_me, has_attachments, has_reactions, and context to add surrounding messages. Messages carry their reactions (tapbacks). Examples: one person's latest messages with {"conversation_query": "dad", "limit": 20}; everything since yesterday grouped by conversation to catch up; a query across all conversations to find when something was discussed. For different wordings, use match=any with several synonyms.
| Argument | Type | Description |
|---|---|---|
| after | string | null | Only messages sent at or after this time: RFC 3339, or YYYY-MM-DD for the start of that day in the person's time zone. |
| limit | integer | null | Messages per page, 1 to 500. Defaults to 100. |
| match | any | all (default): a message must contain every word. any: one word is enough, to search several synonyms in one call. |
| order | any | newest_first (default) or oldest_first. Applies across conversations. |
| query | string | null | Words to find in message text, ignoring case; a word also matches inside longer words, so "relocat" finds "relocating". Up to 8 words. |
| before | string | null | Only messages sent before this time: RFC 3339, or YYYY-MM-DD for the start of that day in the person's time zone. |
| cursor | string | null | The next_cursor from the previous page, with the same other parameters. |
| sender | string | null | Only messages whose sender's name contains this text, ignoring case. |
| context | integer | null | Messages to include before and after each returned message, 0 to 10. Defaults to 0. Useful with a query, to quote a match in context. A page holds at most 1,000 messages counting context, so a large context returns fewer matches per page; follow next_cursor for the rest. |
| from_me | boolean | null | True for only the person's own messages, false for only everyone else's. |
| pattern | string | null | A regular expression the message text must match, ignoring case, for what words cannot express: alternatives, word boundaries, numbers. Example: \\b(flight|plane)s?\\b. Combines with query when both are given. |
| conversations | array | null | Conversation IDs to include, from misento_conversations. Omit for every shared conversation, or pass conversation_query instead. |
| has_reactions | boolean | null | True for only messages someone reacted to (a tapback), false for only messages without reactions. |
| has_attachments | boolean | null | True for only messages with attachments, false for only messages without. |
| response_format | any | concise (default) cuts messages to their first 300 characters; detailed returns them whole. |
| conversation_query | string | null | Instead of conversations: part of a name, phone number, or email, as misento_conversations' query takes it, to read the conversations it finds (up to 10) in one call. `matched` lists them. |
| group_by_conversation | boolean | null | True to return the page grouped per conversation instead of one timeline. |
{
"type": "object",
"$defs": {
"Order": {
"enum": [
"oldest_first",
"newest_first"
],
"type": "string"
},
"QueryMatch": {
"oneOf": [
{
"type": "string",
"const": "all",
"description": "Every word must appear. The default."
},
{
"type": "string",
"const": "any",
"description": "Any one word is enough. Use with several synonyms to search them in one\ncall."
}
],
"description": "How a query's words must appear in a message."
},
"ResponseFormat": {
"oneOf": [
{
"enum": [
"detailed"
],
"type": "string"
},
{
"type": "string",
"const": "concise",
"description": "The default."
}
],
"description": "How much a response carries; each tool says what that means for it."
}
},
"$schema": "https://json-schema.org/draft/2020-12/schema",
"properties": {
"after": {
"type": [
"string",
"null"
],
"description": "Only messages sent at or after this time: RFC 3339, or YYYY-MM-DD for the\nstart of that day in the person's time zone."
},
"limit": {
"type": [
"integer",
"null"
],
"format": "uint32",
"minimum": 0,
"description": "Messages per page, 1 to 500. Defaults to 100."
},
"match": {
"anyOf": [
{
"$ref": "#/$defs/QueryMatch"
},
{
"type": "null"
}
],
"description": "all (default): a message must contain every word. any: one word is\nenough, to search several synonyms in one call."
},
"order": {
"anyOf": [
{
"$ref": "#/$defs/Order"
},
{
"type": "null"
}
],
"description": "newest_first (default) or oldest_first. Applies across conversations."
},
"query": {
"type": [
"string",
"null"
],
"description": "Words to find in message text, ignoring case; a word also matches inside\nlonger words, so \"relocat\" finds \"relocating\". Up to 8 words."
},
"before": {
"type": [
"string",
"null"
],
"description": "Only messages sent before this time: RFC 3339, or YYYY-MM-DD for the start\nof that day in the person's time zone."
},
"cursor": {
"type": [
"string",
"null"
],
"description": "The next_cursor from the previous page, with the same other parameters."
},
"sender": {
"type": [
"string",
"null"
],
"description": "Only messages whose sender's name contains this text, ignoring case."
},
"context": {
"type": [
"integer",
"null"
],
"format": "uint32",
"minimum": 0,
"description": "Messages to include before and after each returned message, 0 to 10.\nDefaults to 0. Useful with a query, to quote a match in context. A page\nholds at most 1,000 messages counting context, so a large context\nreturns fewer matches per page; follow next_cursor for the rest."
},
"from_me": {
"type": [
"boolean",
"null"
],
"description": "True for only the person's own messages, false for only everyone else's."
},
"pattern": {
"type": [
"string",
"null"
],
"description": "A regular expression the message text must match, ignoring case, for\nwhat words cannot express: alternatives, word boundaries, numbers.\nExample: \\\\b(flight|plane)s?\\\\b. Combines with query when both are given."
},
"conversations": {
"type": [
"array",
"null"
],
"items": {
"type": "string"
},
"description": "Conversation IDs to include, from misento_conversations. Omit for every\nshared conversation, or pass conversation_query instead."
},
"has_reactions": {
"type": [
"boolean",
"null"
],
"description": "True for only messages someone reacted to (a tapback), false for only\nmessages without reactions."
},
"has_attachments": {
"type": [
"boolean",
"null"
],
"description": "True for only messages with attachments, false for only messages without."
},
"response_format": {
"anyOf": [
{
"$ref": "#/$defs/ResponseFormat"
},
{
"type": "null"
}
],
"description": "concise (default) cuts messages to their first 300 characters; detailed\nreturns them whole."
},
"conversation_query": {
"type": [
"string",
"null"
],
"description": "Instead of conversations: part of a name, phone number, or email, as\nmisento_conversations' query takes it, to read the conversations it\nfinds (up to 10) in one call. `matched` lists them."
},
"group_by_conversation": {
"type": [
"boolean",
"null"
],
"description": "True to return the page grouped per conversation instead of one timeline."
}
}
}Wait for new messages from others in the shared conversations, instead of polling misento_messages: returns arrived true as soon as any were sent after the given time, or arrived false once timeout_seconds pass. It returns no messages: read them with misento_messages, from_me false and the same after, following next_cursor until it ends, then wait again with the newest sent_at you read. A message shows only once the person's iPhone imported it, which can take minutes.
| Argument | Type | Description |
|---|---|---|
| after* | string | RFC 3339: the newest sent_at you have seen, or when you started waiting. Only messages sent in a later second count. |
| timeout_seconds | integer | null | Optional, 1 to 50: how long to wait. Defaults to 50. |
{
"type": "object",
"$schema": "https://json-schema.org/draft/2020-12/schema",
"required": [
"after"
],
"properties": {
"after": {
"type": "string",
"description": "RFC 3339: the newest sent_at you have seen, or when you started\nwaiting. Only messages sent in a later second count."
},
"timeout_seconds": {
"type": [
"integer",
"null"
],
"format": "uint8",
"maximum": 255,
"minimum": 0,
"description": "Optional, 1 to 50: how long to wait. Defaults to 50."
}
}
}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 17, 2026.