Customermates
REMOTE · CUSTOMERMATES.COM
CRM: Claude, ChatGPT or any MCP client works with records, inbox and company knowledge.
Available components
Last scanned (23 hours ago)
Scores higher than 60% of remote servers
Reachable. Last reached 21 Nov 2026.
Category CRM and Sales
How this component scores in each security and reliability category. Every signal is checked automatically against the live server, and we only credit what we can confirm. How we score → Why this is hard to score →
Endpoint Security94
- The endpoint's TLS certificate is valid, in date, and uses a strong key. View diagnostics → Pass
- Authorisation is enforced on tool calls, advertised via RFC 9728 protected-resource metadata. Discovery is public, which costs nothing: no tool can be invoked without a token. View diagnostics → Pass
- HTTPS is enforced; there's no plaintext access path. View diagnostics → Pass
- The HSTS (Strict-Transport-Security) header is present. View diagnostics → Pass
- DNSSEC check failed: this domain isn't protected by DNSSEC. See how to fix → View diagnostics → Fail
- The authorisation server offers only Dynamic Client Registration (RFC 7591), which MCP 2026-07-28 deprecated in favour of Client ID Metadata Documents. View diagnostics → Partial
Transport & Reachability100
- Verified streamable-http transport via a live MCP handshake. View diagnostics → Pass
Schema Quality & AI Usability53
- AI-judged instruction clarity (good).Pass
- Context-footprint check failed: tool/resource definitions use about 29307 tokens (~465/item across 63 items; 63 tools + 0 resources), over budget; trim descriptions and params. See how to fix → Fail
- Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management10
- Stability observed for 3 of 30 days with no destabilising changes; credit accrues until the full window elapses.Partial
Tool Coverage92
- 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
- 70% of tool parameters carry a description.Partial
- Structured output schemas are declared (100% of tools); any adoption earns full credit.Pass
Tool Safety100
- No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
- All 3 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation.Pass
- An AI judge read all 64 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities60
- Spec-recency check failed: implements MCP spec 2025-06-18; the latest is 2026-07-28. See how to fix → Fail
Checks we ran (13)
- TLS certificate
- Authorization
- HTTPS
- HSTS header
- DNSSEC
- Transport
- Schema quality
- Stability
- Tool coverage
- Injection markers
- Destructive annotations
- Judged manipulation
- MCP protocol
Does not apply to this hosted endpoint (2)
- Supply Chain SecurityScored on package channels only. This is a hosted endpoint, so there is nothing to install or inspect.
- Provenance & TransparencyScored on package channels only. This is a hosted endpoint, so there is nothing to install or inspect.
The score does not measure everything. What the score does not measure →
Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.
- 21 Nov 26 0
- This server's schema is too large to store in full, so we cannot compare its tools day to day functional
- 20 Nov 26 +1
- Schema quality: 25987 → 29307 ▼ functional
- This server's schema is too large to store in full, so we cannot compare its tools day to day functional
- 19 Nov 26 0
- Stability: unverified → 0.03 ▲ functional
- This server's schema is too large to store in full, so we cannot compare its tools day to day functional
- 18 Nov 26 +35
- Transport: unverified → pass ▲ security
- Injection markers: unverified → pass ▲ security
- First check of Judged manipulation: pass security
- Authorization: Authorisation is enforced on tool calls, advertised via RFC 9728 protected-resource metadata. Discovery is public, which costs nothing: no tool can be invoked without a token. security
- MCP protocol: unverified → fail ▼ functional
- Tool coverage: unverified → 100 ▲ functional
- First check of Schema quality: fail functional
- First check of Schema quality: fail functional
- First check of Tool coverage: 72 functional
- First check of Destructive annotations: 100 functional
- First check of Schema quality: good functional
- First check of Tool coverage: 100 functional
- 6 Nov 26 0
- We updated how we score, so this day's move reflects our rubric, not a change to the server See what changed → functional
- 30 Sept 26 0
First indexed and scored.
Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.
Captured 21 Nov 2026 · Probed https://customermates.com/api/v1/mcp
TLS valid
Negotiated TLS 1.3 with TLS_AES_128_GCM_SHA256 .
| Subject | Issuer | Valid from | Valid until | Key | Signature | Serial |
|---|---|---|---|---|---|---|
| CN=*.customermates.com | CN=YR2,O=Let's Encrypt,C=US | 9 Nov 2026 | 7 Feb 2027 | RSA 2048 | SHA256-RSA | 55108e88231b9bfd37b9c5aa8552f83a3ea |
| SANs: *.customermates.com, customermates.com | ||||||
| CN=YR2,O=Let's Encrypt,C=US (CA) | CN=Root YR,O=ISRG,C=US | 3 Sept 2025 | 2 Sept 2028 | RSA 2048 | SHA256-RSA | 4ebd24947e24d394802d84a52fd5b319 |
| CN=Root YR,O=ISRG,C=US (CA) | CN=ISRG Root X1,O=Internet Security Research Group,C=US | 13 May 2026 | 2 Sept 2032 | RSA 4096 | SHA256-RSA | f24b6d17f9d9ad7cb1c9fea78782699f |
Background: What to check on a remote MCP endpoint →
DNSSEC insecure
Validation of customermates.com. — Not signed
| Zone | DS | Keys | Algorithms | Outcome |
|---|---|---|---|---|
| . | Trust anchor | 20326, 38696 | 8, 8 | Verified |
| com. | DS published | 19718 | 13 | Verified |
| customermates.com. | No DS | Unsigned (proven) parent-signed NSEC/NSEC3 proves an unsigned delegation |
Authentication Enforced and verified
The endpoint asked for a token and published valid RFC 9728 metadata describing how to get one.
| Result | Enforced and verified |
|---|---|
| Enforced | On tool calls |
| HTTP status | 200 |
WWW-Authenticate challenge Bearer resource_metadata="https://customermates.com/.well-known/oauth-protected-resource"
Bearer resource_metadata="https://customermates.com/.well-known/oauth-protected-resource" | Header | Value |
|---|---|
| strict-transport-security | max-age=63072000 |
| content-security-policy | frame-ancestors 'self' |
| x-content-type-options | nosniff |
Protected resource metadata
| Document | https://customermates.com/.well-known/oauth-protected-resource |
|---|---|
| Retrieved | Yes |
| Resource | https://customermates.com/api/v1/mcp |
| Authorisation server | https://customermates.com |
Background: How OAuth 2.1 works in the 2026 MCP spec →
Transports 2 probes
| Transport | URL | Outcome | Status | Location |
|---|---|---|---|---|
| streamable-http | https://customermates.com/api/v1/mcp | Verified | 200 | |
| http (plaintext) | http://customermates.com/api/v1/mcp | HTTPS enforced | 308 | https://customermates.com/api/v1/mcp |
The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →
manage_routines Manage routines ~953
Reads, creates, changes and removes routines: saved instructions the assistant runs on a schedule or when a CRM event fires. action list returns compact summaries (id, name, owner, trigger, enabled) by default; includePrompts: true returns full configurations including prompts. owner: 'mine' limits results to routines the caller owns; page, pageSize and searchTerm paginate and search. action runs returns one routine's run history newest first, paginated by cursor, each with its status, summary, triggerKind and error block reason. action create needs name, prompt and triggerKind, plus cronExpression for a schedule, or triggerEvents for an event. Record events (record.created, record.updated, record.deleted) also need recordTrigger for one list or recordSources for several lists; the query filters decide which records fire the routine, and record events can be combined with messaging events in one routine. Omitting enabled on create produces a routine that is immediately LIVE and will start running; enabled false creates a draft. action update changes only the fields supplied, but a change of triggerKind must arrive with that kind's schedule or events or validation fails. action pause disables a routine and settles its queued runs to skipped; re-enabling restores the schedule but not those runs. Only an active system administrator may pause. action run_now starts a scheduled routine immediately. Manual test runs still have triggerKind schedule. It refuses an event routine, another owner's routine, a disabled routine, paused AI work, an inactive owner or exhausted credits before queuing and reports the reason. action delete moves the routine to Trash, where it stops running; the returned trashBatchId identifies the Trash batch it can be restored from, and only deleting it permanently removes its history. A routine whose owner has been deactivated cannot be enabled, and creating one may be refused when the owner has used their plan's per-user routine allowance.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | list = compact routine summaries by default (optional page, pageSize, searchTerm, owner, includePrompts); runs = one routine's run history (id, optional cursor); create = name, prompt, triggerKind, e… |
| cronExpression | – | – | Five-field cron. Runs may be no closer together than 15 minutes. |
| cursor | – | – | runs only. Page cursor returned by a previous runs request. |
| debounceSeconds | integer | – | – |
| enabled | boolean | – | Omitting this on create produces a LIVE routine; false creates a draft. |
| expectedSchemaRevision | integer | – | Required when supplying recordTrigger: the current schema revision of the record model, as returned by type discovery. |
| id | string | – | Routine id. Required for update, pause, run_now and delete. |
| includePrompts | boolean | – | list only. Include the full routine configurations and prompts; false returns compact summaries. |
| name | string | – | Required on create. |
| owner | string | – | list only. mine returns routines owned by the caller. |
| page | integer | – | 1-indexed page number |
| pageSize | integer | – | list only. Results per page, any whole number from 1 to 100, served exactly: page 2 of pageSize 50 holds records 51 to 100. When a result was truncated, ask again with about half the size. Default 25. |
| prompt | string | – | The instructions the assistant receives on every run. Required on create. |
| recordSources | – | – | For record events from several lists, or with different filters per event: each source has its own query (typeId, filters, relationships), changedFieldIds and the record events it handles. Every reco… |
| recordTrigger | – | – | For record.created/updated/deleted from one list: a query with stable typeId, filters and relationships, plus watched field IDs (changedFieldIds). The query filters decide which records fire the rout… |
| searchTerm | string | – | list only. Free-text match against the routine name. |
| timezone | – | – | IANA zone the cron is read in. |
| triggerEvents | array | – | Events an event routine reacts to. Required for triggerKind event. |
| triggerKind | string | – | Required on create. |
| Name | Type | Req | Description |
|---|---|---|---|
| deleted | boolean | – | – |
| enabled | boolean | – | – |
| id | string | – | – |
| items | array | – | – |
| name | string | – | – |
| nextCursor | – | – | – |
| page | number | – | – |
| pageSize | number | – | – |
| started | boolean | – | – |
| total | number | – | – |
No examples provided.
manage_rules Manage rules ~932
Reads, creates, changes and removes automation rules: deterministic when-if-then rules that run without AI, on every plan and on self-hosted instances. A rule has a trigger (a record created or updated, a field changed to a value, a date reached with an offset in days before or after, a message received, a form submitted, an incoming webhook or a booking cancelled or rescheduled), optional conditions on the triggering record and one or more effects (set a field, create a record such as an Action, notify members, send an outbound webhook). A rule runs once per record and occurrence, as its owner, with the owner's permissions: it cannot do more than the owner could. Changes a rule makes cannot start the same rule again, and a chain stops after two rules. action list returns compact summaries; get returns one rule in full; runs returns one rule's history, newest first, each with its status, the affected record and the outcome of every effect; a run whose record the caller may not read has recordId null, and a date rule's run then also scheduledFor null. action create needs name, trigger and effects; omitting enabled produces a LIVE rule and enabled false creates a draft. Plans limit the number of active rules; the refusal names the limit. action update changes only the keys supplied; trigger, conditions and effects REPLACE their previous value. Only the owner may change or enable a rule. action enable and disable switch a rule on and off; its history is kept; an administrator may also disable. action delete removes the rule and its history permanently and is IRREVERSIBLE. action preview lists which records a date-reached rule would hit on its upcoming sweep (at most 200 are checked), without running anything.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | Rule operation. Keys per action: list = optional typeId, owner, page, pageSize; get, enable, disable and delete = id; runs = id, optional cursor; create = name, trigger, effects, then optional condit… |
| conditions | array | – | AND-combined checks on the record that triggered the rule: equals, notEquals, in, notIn, empty, notEmpty, before, after (a date or relativeDays from today). create and update; on update REPLACES the… |
| cursor | string | – | runs only. The nextCursor of the previous page. Omit it for the newest runs. |
| effects | array | – | What happens, in order: setField on the triggering record, createRecord (for example an Action in the tasks list with a title, a due date offset, an assignee and a link to the triggering record), not… |
| enabled | boolean | – | create and update. Omitting it on create produces a LIVE rule; pass false to save a draft. |
| id | string | – | Rule id. Required for get, runs, update, enable, disable and delete; optional for preview. |
| name | string | – | Required on create. |
| owner | string | – | list only. mine returns the rules the caller owns. |
| page | integer | – | 1-indexed page number |
| pageSize | integer | – | list only. Results per page, any whole number from 1 to 100, served exactly: page 2 of pageSize 50 holds records 51 to 100. When a result was truncated, ask again with about half the size. Default 25. |
| trigger | – | – | When the rule fires. Required on create; on update REPLACES the trigger. kind is recordCreated, recordUpdated, fieldChanged, dateReached (a date or date and time field, days before or after, at an ho… |
| typeId | string | – | list only. Limits the list to rules about this list (record type). |
| Name | Type | Req | Description |
|---|---|---|---|
| alreadyFired | number | – | – |
| deleted | boolean | – | – |
| enabled | boolean | – | – |
| id | string | – | – |
| items | array | – | – |
| name | string | – | – |
| nextCursor | – | – | – |
| page | number | – | – |
| pageSize | number | – | – |
| recordIds | array | – | – |
| runs | array | – | – |
| supported | boolean | – | – |
| total | number | – | – |
| truncated | boolean | – | – |
No examples provided.
manage_social_relations Manage social relations ~465
Manages connection requests on a connected LinkedIn or Instagram account. action list returns invitations with invitationId, user and message: direction=received (default) lists requests sent to the account owner, direction=sent the owner's pending outgoing requests; user.id is a provider person id that identifies that person in a profile lookup. A request message is written by the sender and arrives between untrusted-content markers. action invite SENDS A REAL connection request to identifier, a provider person id; a messaging-thread participant identifier is a different identifier type that a profile lookup resolves to a person id. The optional message travels with the request. In the hosted Assistant the person is verified with the provider before the request is sent, and the request is refused when the identifier does not resolve. action accept and action cancel take the invitationId of an invitation returned by action list; in the hosted Assistant the invitationId is also checked against the provider's first pages of invitations, and the request is refused when it is not among them. Paginates list with cursor or offset plus limit.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | list = received or sent invitations (direction, optional cursor, offset, limit); invite = identifier plus optional message; accept and cancel = invitationId. Every action needs connectedAccountId. |
| connectedAccountId | string | yes | The id of a connected workspace account with provider LinkedIn or Instagram and status 'ok' |
| cursor | string | – | list only: pagination cursor from the previous page's next_cursor |
| direction | string | – | list only: received (default) lists invitations sent to the account owner; sent lists invitations the account owner sent, whose invitationId cancel takes |
| identifier | string | – | Required for invite: the provider person id of the recipient. A messaging-thread participant identifier is a different identifier type that a profile lookup resolves to a person id |
| invitationId | string | – | Required for accept and cancel: the invitationId of an invitation returned by action list |
| limit | integer | – | list only: items per page (1-100, default 10) |
| message | string | – | Optional note sent with an invite, at most 300 characters |
| offset | integer | – | list only: pagination offset, for results without a cursor |
| Name | Type | Req | Description |
|---|---|---|---|
| invitationId | – | – | – |
| items | array | – | – |
| next_cursor | – | – | – |
| status | string | – | – |
| total | number | – | The provider's count of every matching row, present only when the provider reports one |
No examples provided.
manage_team Manage team ~593
Administers team members. action invite SENDS REAL INVITATION EMAILS to the given addresses (up to 20 per request). action invite_link returns the workspace invitation link and its expiresAt, as the Invite modal shows it, creating one valid for 7 days when none is active; anyone who opens it can request to join. action update_member changes an existing member's role or status: the member is identified by userId or email (both must match if supplied), plus roleId and/or status; omitted fields keep their current values; a member who has no role assigned yet (e.g. a still pending invite) has no role to keep, so roleId is required with the change and the request is rejected without it. action remove_member removes another member from this workspace (userId or email, both must match if supplied): their status becomes inactive and they show as left (a declined request leaves the list rather than showing as left), it permanently revokes their API keys and MCP connections for this workspace, disconnects the mailboxes and LinkedIn accounts they connected here, pauses their routines and frees an active member's seat; for a member waiting for approval it declines the request. It SENDS A REAL EMAIL telling the removed member that their access ended (none for a declined request). Records they own stay. To return they ask to join again with an invitation link, and an admin has to approve the request; the caller cannot remove themselves, the only active membership of the last Customermates platform operator, a member who already left or the last active admin. last-admin protection is enforced server-side, so the workspace cannot lose its last active admin. invite requires create permission on users, as does invite_link; update_member requires update permission on users (Edit on Users & Roles), the same check as REST POST /v1/members/update, and remove_member requires the same permission as REST POST /v1/members/remove.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | invite = send invitation emails (emails); invite_link = return the workspace invitation link for the caller to share; update_member = change an existing member's role or status (userId or email plus… |
| string | – | update_member and remove_member actions: member email; provide email or userId, and both must match when supplied | |
| emails | array | – | invite action: required, 1 to 20 email addresses, each receives a real invitation email |
| roleId | string | – | update_member action: id of the new role. Omit to keep the current role. |
| status | string | – | update_member action: new status. Omit to keep the current status. |
| userId | string | – | update_member and remove_member actions: user id of the member; provide userId or email |
| Name | Type | Req | Description |
|---|---|---|---|
| string | – | – | |
| expiresAt | string | – | – |
| failed | number | – | – |
| failedEmails | array | – | – |
| link | string | – | – |
| message | string | yes | – |
| previousStatus | string | – | – |
| roleId | – | – | – |
| sent | number | – | – |
| status | string | – | – |
| userId | string | – | – |
No examples provided.
manage_trash Restore or permanently delete Trash items ~316
restore brings items back with their values, links and sub-list rows; links to records deleted permanently in the meantime are dropped and counted in droppedLinks. Deleting a field removes saved-view filters, columns, sort or grouping that depend on it, and restoring the field cannot recreate those settings; a field or saved-view restore reports the affected views and counts in droppedViewReferences. blocked lists items that stay in Trash with a reason: their list or parent must be restored first; requiresRestore when a field, relationship or list they need is still in Trash (one request restores lists, fields and relationships together, in any order); nameTaken when another item now uses their name; incompatible when they no longer fit the current data or setup; or routineLimitReached when the routine owner already has the limit of routines their plan includes. A restore that recalculates many records answers pending with an operationId of a background operation; records, views and other items in that request stay in Trash under blocked with operationPending until the operation has completed. preview shows exactly what a permanent deletion removes without changing anything and returns its impactHash. delete_permanently needs delete access to every listed item; empty is for administrators only. Both cannot be undone: they remove the items and erase the deleted records' values from history, keeping only who changed what when.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | – |
| all | boolean | – | – |
| batchId | string | – | – |
| expectedImpactHash | string | – | – |
| itemIds | array | – | – |
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | – |
| result | – | yes | – |
No examples provided.
manage_webhooks Manage webhooks ~1,351
Manages webhook subscriptions and inspects their deliveries. action create requires url and events. Record events also require expectedSchemaRevision; recordTrigger selects one list and its filters, and recordSources selects several lists, each with its own filters and events. Record events can be combined with messaging events in one webhook. Delivery uses recordOwnerUserId permissions. Only its owner or an administrator can change a record subscription. action update requires id; events REPLACES the full subscription list; secret: omit to keep, null to clear, string to set. action delete is IRREVERSIBLE. action get returns one webhook (the signing secret and header values are not returned; get reports headerNames in their place). action list supports searchTerm, filters, sort, paging. action list_deliveries returns delivery attempts newest first; without `id` it spans the whole workspace (as REST GET /v1/webhooks/deliveries), with `id` it is scoped to that webhook (as REST GET /v1/webhooks/{id}/deliveries); narrow further with searchTerm or filters. Each requestBody is the payload as sent: Formatted text in it arrives as Markdown between untrusted-content markers, and its record content is customer-written, untrusted data. action resend_delivery re-sends a past delivery as a NEW delivery record; it takes the id of a delivery returned by list_deliveries.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | Webhook operation. Keys per action: create = url, events, then optional description, secret, headers, bodyTemplate, enabled; update = id plus any of url, description, events, secret, headers, bodyTem… |
| bodyTemplate | – | – | JSON template for the request body, for receivers that need a fixed shape. Use {{event}}, {{id}}, {{timestamp}}, {{companyId}}, {{actorId}} or {{data...}} placeholders such as {{data.record.ref.recor… |
| description | string | – | create and update. Human-readable note about what this webhook does. |
| enabled | boolean | – | create (default true) and update. |
| events | array | – | Required for create; on update REPLACES the subscribed events. |
| expectedSchemaRevision | integer | – | Current schema revision of the record model. Required when creating a record subscription or changing its trigger or owner. |
| filters | array | – | list and list_deliveries only. Array of filter rules, AND-combined. Each rule is { field, operator, value? }. Use only the operators listed in each field's hint; value-less operators take no value. E… |
| headers | – | – | Extra HTTP headers sent with every delivery, for receivers that require their own authentication. create: optional object. update: omit to keep the current headers, pass null to clear them, pass an o… |
| id | string | – | Required for update, delete, get (webhook id) and for resend_delivery (delivery id from list_deliveries). Optional for list_deliveries (webhook id): scopes deliveries to that webhook, including those… |
| page | integer | – | 1-indexed page number |
| pageSize | integer | – | Results per page, any whole number from 1 to 100, served exactly: page 2 of pageSize 50 holds records 51 to 100. When a result was truncated, ask again with about half the size. Default 25 for list a… |
| recordOwnerUserId | string | – | Member whose current record access governs delivery. Defaults to the creator. Only an administrator may choose another member; ownership does not grant record access. |
| recordSources | – | – | For record events from several lists, or with different filters per event: each source has its own query (typeId, filters, relationships), changedFieldIds and the record events it handles. Every reco… |
| recordTrigger | – | – | For record.created, record.updated and record.deleted from one list: a stable type ID, query filters and watched field IDs. Null selects every accessible type. On update omit to keep the current trig… |
| searchTerm | string | – | list matches url and description; list_deliveries matches url and event name. |
| secret | – | – | Shared secret used to sign outgoing requests. create: optional string. update: omit to keep the current secret, pass null to clear it, pass a string to set a new one. |
| sortDescriptor | object | – | list and list_deliveries only. Sort by one field: { field, direction: "asc" | "desc" }. Sortable fields: list: name, createdAt, updatedAt; list_deliveries: createdAt. |
| url | string | – | Endpoint that will receive event POST requests (https recommended). Required for create. |
| Name | Type | Req | Description |
|---|---|---|---|
| deleted | boolean | – | – |
| events | array | – | – |
| id | string | – | – |
| items | array | – | – |
| newDeliveryId | string | – | – |
| page | number | – | – |
| pageSize | number | – | – |
| resentDeliveryId | string | – | – |
| total | number | – | – |
| url | string | – | – |
No examples provided.
manage_widgets Manage widgets ~1,152
Reads, creates, changes and deletes dashboard widgets. A created widget stays on the dashboard. list returns IDs, names, each widget's dashboard view (viewId, null for the main dashboard) and its current layout within that view; get (ids) returns saved configuration and access-filtered results. A chart is created with name, measure, displayOptions, expectedRevision and idempotencyKey. displayOptions.displayType must fit the measure: number needs groupBy null, areaChart needs groupBy.dateInterval, rankedTable needs a grouping and funnelChart needs a single-choice grouping field; other chart styles accept any grouping. A chart is updated with id, expectedVersion, expectedRevision, idempotencyKey and changed fields; measure and displayOptions replace the complete previous value. Results preserve exact decimal strings and missing, restricted or error states. Activity timelines use kind activityTimeline, name, activityQuery, expectedRevision, idempotencyKey and showFilters. activityQuery scopes stable record/type IDs and combines typed inclusion/exclusion filters; updates also require expectedVersion. Calendar widgets use kind calendar and calendarQuery: layout upcoming lists the records of the coming days (days, today included) with an overdue section for earlier dates (showOverdue) and an optional amount column; layout month shows each day of the current month as a heat grid with a count, or the sum of amountFieldId with aggregation sum. dateFieldId must be a date, date-time, date range or date-time range field of the source list; a range puts each record on the day it starts and counts as overdue once it has ended; timeZone sets today and the day of date-time values. get with a calendar widget also accepts window {from, to} (at most 42 days) to read another month or one day. Map widgets use kind map and mapQuery: each record appears at the location stored with its address field (the building after an address lookup, otherwise the centre of its postal code), colorFiel…
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | create: name, expectedRevision, idempotencyKey; update: id, expectedVersion, expectedRevision, idempotencyKey; get: ids; delete: id. list requires no other fields. Create also requires the chosen kin… |
| activityQuery | object | – | – |
| calendarQuery | object | – | Calendar widgets: the records (source), the date field that puts them on days, layout upcoming or month and the IANA time zone. |
| displayOptions | object | – | – |
| expectedRevision | integer | – | Required for create and update: the current record model configuration revision. |
| expectedVersion | integer | – | Required for update: the saved widget version as returned by get or list; not accepted on create. |
| id | string | – | – |
| idempotencyKey | string | – | Required for create and update: a stable retry key; the same request repeats the same key and a different change takes a new key. |
| ids | array | – | – |
| isTemplate | boolean | – | – |
| kind | string | – | Widget kind on create: chart (default), activityTimeline, calendar or map. A saved widget keeps its kind. |
| layout | object | – | Optional position and size on the 12-column desktop grid. Omit it to keep the current position on update or to place a new widget in the first free spot. It must not overlap another widget. |
| mapQuery | object | – | Map widgets: the records (source) and the address field whose stored location places them, with optional color, size and detail fields. |
| measure | object | – | Count or aggregate records at source.typeId. Field and relationship references are stable IDs. Group paths and assignee groups use full attribution; the separate overall total counts each source reco… |
| name | string | – | – |
| showFilters | boolean | – | Activity timeline, calendar and map appearance: show the filter summary on the card. Chart options belong in displayOptions. |
| viewId | – | – | Dashboard view (a saved view of the dashboard surface) that holds the widget, or null for the main dashboard. Omitted: the caller's current dashboard view on create, the widget's own view on update. |
| window | object | – | get only, calendar widgets: the days to read in place of the widget's own range. |
| Name | Type | Req | Description |
|---|---|---|---|
| deleted | boolean | – | – |
| id | string | – | – |
| items | array | – | – |
| kind | string | – | – |
| layout | object | – | – |
| name | string | – | – |
| version | number | – | – |
No examples provided.
manage_wiki_pages Manage Knowledge Base pages ~1,005
Knowledge Base of company facts, processes, voice and support guidance. kind: guide = the one Operating Guide; procedure = numbered steps + whenToUse; default knowledge. list: 5 per page, or pageSize up to 100, guide pinned first, remaining pages in saved order; chars is the length of a page's Markdown. search: hasMore marks further pages; total is a lower bound unless totalIsExact. Returns snippets, section offsets and section anchors, plus didYouMean when a misspelled word was corrected. get: one Markdown chunk of up to 5,500 characters, or maxChars up to 20,000 (outline at 0, each heading with its section anchor); nextOffset is the offset of the following chunk and null at the end; updatedAt changes when the page changed between chunks; anchor reads one section. An anchor is unique within its page: a repeated heading takes -2, -3 and so on. The first chunk names up to 8 pages that link to this page (linkedFrom, linkedFromTotal); linkedFromIncomplete and, on links, syncing mark a list that is still being read. owner and verifiedAt appear when set; verifiedAt is when someone last confirmed the page, updatedAt when its text last changed. links: with id, the pages that link to the page (linkedFrom), the pages it links to (linksTo), linked page ids that no longer exist (missingPageIds) and the record lists it links into; without id, the pages no other page links to (orphans) and the pages with links to deleted pages (broken), 50 each with totals. update: edits change parts of a page (find and replace one exact text, replace or insert beside a section by its anchor, or add at the start or end) without resending the whole markdown; the result carries totalChars, and a page holds at most 65,535 characters. move: places page id before or after targetId in the saved order; the Operating Guide stays pinned first, so it can neither move nor serve as targetId. gaps: questions the Knowledge Base could not answer (editors). delete moves the page to Trash, where it is deleted p…
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | list: page, pageSize, kind; search: query, page; get: id, offset or anchor, maxChars; links: id; create: pages, requireEmpty; update: id, expectedUpdatedAt, title, markdown, edits, kind, whenToUse, o… |
| anchor | string | – | Section anchor as returned by a search hit or the outline; get returns only that section. |
| edits | array | – | update only, in place of markdown: changes to parts of the page, applied in order and all or none, so a long page is not resent. Each edit is one of: find with replace; anchor with markdown; markdown… |
| expectedUpdatedAt | string | – | updatedAt from a prior read. |
| id | string | – | Page id. |
| kind | string | – | – |
| markdown | string | – | Markdown with actual newline characters; escaped backslash-n separators are rejected. |
| maxChars | integer | – | get only: the most characters one response holds, 2500 to 20000. Default 5500; a larger value reads a long page in fewer chunks. |
| offset | integer | – | Hit offset or prior nextOffset. |
| owner | – | – | update only: who keeps the page current, a person, a team or an agent; null clears it. |
| page | integer | – | 1-indexed page number |
| pageSize | integer | – | list only: pages per response, 1 to 100. Default 5; 100 returns the inventory of most Knowledge Bases at once. |
| pages | array | – | Created atomically. |
| placement | string | – | move only: before or after targetId in the saved order. |
| query | string | – | Search terms. |
| requireEmpty | boolean | – | Only into an empty Knowledge Base. |
| targetId | string | – | move only: the reference page this page is placed beside. |
| title | string | – | New title. |
| verified | boolean | – | update only: true records that the page was checked against its sources and is still correct. verifiedAt becomes now; updatedAt moves only when the content changes. |
| whenToUse | string | – | – |
| Name | Type | Req | Description |
|---|---|---|---|
| broken | array | – | – |
| brokenTotal | number | – | – |
| deleted | boolean | – | – |
| didYouMean | array | – | – |
| gaps | array | – | – |
| gapsHandling | string | – | – |
| gapsTrust | string | – | – |
| hasMore | boolean | – | – |
| hint | string | – | – |
| id | string | – | – |
| incomplete | boolean | – | – |
| items | array | – | – |
| linkedFrom | array | – | Pages that link to this page; a rename, split or delete of the page affects them. |
| linkedFromIncomplete | boolean | – | The pages that link to this page could not all be read yet, so linkedFrom may miss some. |
| linkedFromTotal | number | – | – |
| links | array | – | – |
| linksTo | array | – | – |
| linksTruncated | boolean | – | – |
| markdown | string | – | – |
| markdownChunk | string | – | – |
| missingPageIds | array | – | Linked page ids that no longer name a live page. |
| moved | boolean | – | – |
| nextOffset | – | – | – |
| noAdequateSection | boolean | – | – |
| offset | number | – | – |
| orphans | array | – | – |
| orphansTotal | number | – | – |
| outline | array | – | – |
| outlineTruncated | boolean | – | Some headings were omitted to keep the response bounded. |
| page | number | – | – |
| pageSize | number | – | – |
| pages | number | – | – |
| records | array | – | – |
| retrieval | string | – | – |
| section | string | – | – |
| syncing | boolean | – | – |
| title | string | – | – |
| total | number | – | – |
| totalChars | number | – | – |
| totalIsExact | boolean | – | – |
| url | string | – | – |
No examples provided.
move_email_thread Move email thread to a folder ~190
Moves eligible emails to another folder in the provider mailbox. Requires threadId and folderId, where folderId is the id of a target in the thread's folder.moveTargets, not a folder name. An optional messageId from that thread moves only that email. Sent and Drafts stay in place and are not move targets. Emails moved outside the watched folders disappear from the inbox; other visible emails keep the conversation available. Returns the folder name and the moved, skipped and failed counts, and reports a provider rate limit or an early stop with its reason.
| Name | Type | Req | Description |
|---|---|---|---|
| folderId | string | yes | Id of the target folder, taken from the thread's folder.moveTargets; a folder name is not accepted |
| messageId | string | – | Email message id in this thread. Omit to move all eligible emails |
| threadId | string | yes | Id of the email thread, as listed in the inbox |
| Name | Type | Req | Description |
|---|---|---|---|
| failedCount | number | yes | – |
| folderId | string | yes | – |
| folderName | string | yes | – |
| hiddenFromInbox | boolean | yes | – |
| movedCount | number | yes | – |
| rateLimited | boolean | yes | – |
| retryAfter | string | – | – |
| skippedCount | number | yes | – |
| stoppedMessage | string | – | Readable explanation of stoppedReason |
| stoppedReason | string | – | Error code when the provider stopped the move part way for a reason other than a rate limit |
| threadId | string | yes | – |
No examples provided.
mutate_crm_record Change a record or relationship ~690
Creates, updates, deletes, links or unlinks records. updateMany applies one shared patch atomically to a typed target array; deleteMany previews and deletes the whole selection atomically. Each target carries its latest version. Writes use field-assignment arrays and stable references. Omitted fields remain unchanged; null explicitly clears an optional input. A full Formatted text replacement writes {kind: 'richText', documentJson: '<ProseMirror document JSON string>'}; Markdown is returned on read but is not accepted for full replacement. A file value lists the ids of uploaded files in order and replaces the files in the field, so the ids to keep are included; an address value sends its parts and is stored unverified unless it is {kind: 'address', candidateToken} from an address lookup. To add Markdown to a Formatted text or single Text field without rewriting it, an update sends { fieldId, append }: the text is added after the current value with a blank line, and an update that only appends cannot conflict with a newer record version. Phone numbers, email addresses and social profiles are the record's identities, sent as mutation.identities (provider and value) on create or update, with fields: [] when nothing else changes; the list replaces the contact details, so the existing ones to keep are included. Calculated fields cannot be written. An idempotency key identifies one exact payload, and a retry of that payload reuses it. A version conflict is rejected with the current version of each target, which is the expectedVersion for a retry. delete and deleteMany move the records, their sub-list rows and cascading records to Trash, where they stay restorable for 30 days; the result carries trashBatchId identifying the whole delete. permanent: true deletes permanently right away (erasure requests) and cannot be undone. The impactHash of a deletion preview, passed as expectedImpactHash, rejects the deletion when its cascading effects changed. Pending operations pause…
| Name | Type | Req | Description |
|---|---|---|---|
| expectedRevision | integer | yes | – |
| idempotencyKey | string | yes | – |
| mutation | object | yes | create requires typeId and fields; update requires ref, expectedVersion and fields, where a field entry is { fieldId, value } or, for Formatted text and Text fields, { fieldId, append } to add text a… |
| Name | Type | Req | Description |
|---|---|---|---|
| result | – | yes | – |
No examples provided.
preview_crm_deletion Preview deletion ~187
Returns a preview of the records and links a deletion removes, including cascading line items; no records are changed. The request takes expectedRevision plus either a single ref with expectedVersion, or targets for an atomic selection deletion. The result lists the calculation definitions that may need recalculation and carries an impactHash identifying the current cascading effects, which a deletion accepts as expectedImpactHash; a null removedLinks means the total is restricted. Both preview and deletion enforce record access, deletion policies and protected capabilities.
| Name | Type | Req | Description |
|---|---|---|---|
| expectedRevision | integer | yes | The current data model revision, as reported with the record types and the record model. |
| expectedVersion | integer | – | The latest version of ref. |
| ref | object | – | The record to delete. Requires expectedVersion and excludes targets. |
| targets | array | – | The selection for an atomic deleteMany preview. Excludes ref and expectedVersion. |
| Name | Type | Req | Description |
|---|---|---|---|
| calculations | array | yes | – |
| impactHash | string | yes | – |
| recordVersion | integer | – | – |
| ref | object | – | – |
| removedLinks | – | yes | – |
| removedRecords | array | yes | – |
| schemaRevision | integer | yes | – |
| targets | array | – | – |
No examples provided.
publish_social_post Publish social post ~539
Publishes a post from a connected LinkedIn or Instagram account. SIDE EFFECT: publishes a real post that this tool cannot take back. connectedAccountId is a connected workspace account with provider linkedin or instagram and status ok, owned by the caller or shared. fileIds attach files from the file field of a record readable by the caller: several images become a carousel, otherwise one video or one document. visibility relations_only limits a LinkedIn post to the account's connections. postAs publishes as a LinkedIn company page the account administers. quotedPostId reposts that LinkedIn post: empty text is a plain repost, text makes it a quote repost. firstComment is posted on the new post right after it is published; when that fails the post stays published and comment_error names the reason. Returns id (the post id), share_url, created_at, comment_id, comment_error and repeated. An idempotencyKey makes a retry safe: the same request with the same key returns the first result. A failed publish with an unknown outcome may already have gone out.
| Name | Type | Req | Description |
|---|---|---|---|
| connectedAccountId | string | yes | The id of a connected workspace account with provider LinkedIn or Instagram and status 'ok' |
| fileIds | array | – | Optional files to attach, in order: file ids from a file value or from the files list of a record readable by the caller. Several images form a carousel; otherwise one video or one document. Together… |
| firstComment | string | – | Optional comment the account posts on its own post right after publishing (at most 1250 characters) |
| idempotencyKey | string | – | Optional UUID. Repeating the identical request with the same key returns the first result with repeated true rather than publishing again; a different payload with a used key is refused |
| postAs | string | – | LinkedIn only: the id of a company page the account administers, to publish as that page rather than as the person; the first comment is posted as the page too |
| quotedPostId | string | – | LinkedIn only: the id of the post to repost, as listed in a posts result. With empty text it is a plain repost; with text it is a quote repost |
| text | string | yes | The post text, published exactly as given (LinkedIn at most 3000 characters, Instagram captions 2200). May be empty when fileIds or quotedPostId is set |
| visibility | string | – | LinkedIn only: anyone (provider default) or relations_only (the account's connections) |
| Name | Type | Req | Description |
|---|---|---|---|
| comment_error | – | yes | – |
| comment_id | – | yes | – |
| created_at | – | yes | – |
| id | string | yes | – |
| repeated | boolean | yes | – |
| share_url | – | yes | – |
No examples provided.
query_crm_measure Calculate a report measure ~180
Aggregates one contribution per source record at an explicit grain. Grouping across relationships, by system:assignedTo or by a multiple choice field uses full attribution, so group totals can exceed the distinct overall total. A time series groups by a date or dateTime field, system:createdAt or system:updatedAt with groupBy.dateInterval (day, week, month, quarter or year; ISO weeks start Monday) and an IANA groupBy.timeZone; only periods with records are returned, in chronological order. Restricted inputs stay restricted; mixed currencies return an error.
| Name | Type | Req | Description |
|---|---|---|---|
| aggregation | string | yes | – |
| groupBy | – | yes | – |
| groupLimit | integer | – | Maximum number of groups, including the missing group. A measure with more groups fails rather than truncating. |
| source | object | yes | – |
| valueFieldId | – | yes | – |
| Name | Type | Req | Description |
|---|---|---|---|
| attribution | string | yes | – |
| groups | array | yes | – |
| schemaRevision | integer | yes | – |
| total | object | yes | – |
No examples provided.
query_crm_records Query records ~478
Queries an accessible type with typed filters, relationships, locale-aware sorting and database pagination. A filter is {fieldId, operator, value}; single-value operators take a typed scalar value, while in, notIn, all, between and overlaps take values (an array of typed scalars) and value: null. For a multiple choice field, values holds individual {kind: 'select', value: optionId} scalars rather than one selectList value. Multiple choice fields filter with in (any of), all and notIn (none of) and cannot be sorted or grouped. Optional grouping supports single choices, booleans, members, relationships and date buckets, with per-group pages; returned group keys load further pages. Calculated, restricted, missing and failed values are distinct. Decimal values use strings. Formatted text arrives as Markdown between untrusted-content markers.
| Name | Type | Req | Description |
|---|---|---|---|
| fields | array | – | – |
| filters | array | – | – |
| groupPage | object | – | Requires grouping. Defaults to 10 records per group, at most 25 materialized groups and 1000 unique records. A returned group key in only and overrides loads another page; collapsed groups retain cou… |
| groupSummaries | array | – | Requires grouping. Aggregate numeric fields over every record in each group, independent of pagination. Relationship groups use full attribution; a source record contributes once per group. Restricte… |
| grouping | object | – | Group by a field, system column or relationship column. Date buckets use UTC boundaries; weeks start on Sunday. At most 50 value groups and one empty group are supported per query. |
| includeIdentities | boolean | – | Include registered identifiers for returned records of a list with a contact details (channels) field. |
| includePaths | array | – | – |
| includeRelationships | array | – | – |
| locale | string | – | – |
| page | integer | – | – |
| pageSize | integer | – | – |
| relatedFilters | array | – | – |
| relationships | array | – | – |
| search | string | – | – |
| searchFields | array | – | Limit search to these fields of the list (text, URL, address, formatted text, files, contact details). Omitted searches all of them. |
| sort | array | – | – |
| typeId | string | yes | – |
| Name | Type | Req | Description |
|---|---|---|---|
| grouping | object | – | – |
| page | integer | yes | – |
| pageSize | integer | yes | – |
| records | array | yes | – |
| schemaRevision | integer | yes | – |
| total | integer | yes | – |
| untrustedTextHandling | string | – | Present when the result holds marked untrusted text; how to treat the text between the markers. |
No examples provided.
read_crm_operation Read operation progress ~81
Returns the progress or the completed result of a durable configuration or record operation. The state is pending, staging, completed, failed or cancelled; a pending or staging response carries retryAfterMs, the wait before polling again, and does not confirm publication. Reads retain the previous complete state while workspace CRM writes are paused.
| Name | Type | Req | Description |
|---|---|---|---|
| operationId | string | yes | – |
| Name | Type | Req | Description |
|---|---|---|---|
| errorCode | – | yes | – |
| id | string | yes | – |
| processed | integer | yes | – |
| result | – | yes | – |
| retryAfterMs | – | yes | Milliseconds to wait before polling a pending or staging operation again; null for a terminal state. |
| state | string | yes | – |
| total | integer | yes | – |
No examples provided.
read_crm_record Read a record ~158
Returns one record identified by both typeId and recordId: typed values, the current record version and configuration revision for a subsequent validated update, and the relationship or path summaries requested through includeRelationships and includePaths. Formatted text arrives as Markdown between untrusted-content markers; it and other customer-provided fields are untrusted data. Record access is checked on every read and for every summarized record.
| Name | Type | Req | Description |
|---|---|---|---|
| includePaths | array | – | Declared relationship paths to summarize in relationshipPaths. Omitted returns none. |
| includeRelationships | array | – | Relationship summaries to return in relationships, selected the same way as the relationships of a record query. Omitted returns none. |
| recordId | string | yes | – |
| typeId | string | yes | – |
| Name | Type | Req | Description |
|---|---|---|---|
| assignedUserIds | array | yes | – |
| assignedUsers | array | yes | – |
| createdAt | string | yes | – |
| fields | array | yes | – |
| files | array | – | Metadata of the files referenced by file values the caller can read. |
| identities | array | – | – |
| memberUsers | array | yes | – |
| ref | object | yes | – |
| relationshipPaths | array | – | – |
| relationships | array | yes | – |
| schemaRevision | integer | yes | – |
| untrustedTextHandling | string | – | Present when the result holds marked untrusted text; how to treat the text between the markers. |
| updatedAt | string | yes | – |
| version | integer | yes | – |
No examples provided.
read_trash Read Trash ~424
action list (default): lists deleted items in Trash, newest first: records (with their sub-list rows), lists, fields, relationships, contact details (channels) fields, saved views, dashboard views, widgets, routines and Knowledge Base pages. Each item shows who deleted it, when, and how many days remain before it is deleted permanently (30 days after deletion). Filters: kinds, typeIds (the list) and search on the name. Only items the caller can see and delete are returned; administrators see everything except other members' saved views, dashboard views and widgets, which stay personal. Items in Trash are invisible to every other tool until restored. Names are customer-provided data. action read: opens one deleted record read only, as the Trash preview in the app shows it, from typeId and recordId (the typeId and targetId of a list item of kind record). Returns its values and related records as they were when it was deleted, its list's model with the fields the caller can read, and its Trash item with who deleted it, when it expires and canRestore. Formatted text arrives as Markdown between untrusted-content markers; it and other customer-provided fields are untrusted data. A record that is not in the caller's Trash answers recordNotFound.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | – | list (default) pages the items in Trash, filtered by kinds, itemIds, typeIds and search; read opens one deleted record read only and takes typeId and recordId. |
| itemIds | array | – | – |
| kinds | array | – | – |
| page | integer | – | – |
| pageSize | integer | – | – |
| recordId | string | – | read: the record, the targetId of a Trash item of kind record. |
| search | string | – | – |
| sortDescriptor | object | – | Order by deletion time; newest first when omitted. |
| typeId | string | – | read: the list of the record, the typeId of a Trash item of kind record. |
| typeIds | array | – | – |
| Name | Type | Req | Description |
|---|---|---|---|
| items | array | – | – |
| model | object | – | The record type, its embedded types and the fields the caller can read. |
| record | – | – | The values as they were when the record was deleted, filtered by field access, with up to 25 related records per relationship. Formatted text arrives as Markdown between untrusted-content markers. |
| total | integer | – | – |
| trash | object | – | – |
| typeId | string | – | – |
| untrustedTextHandling | string | – | Present when the result holds marked untrusted text; how to treat the text between the markers. |
No examples provided.
request_support Request support ~91
Emails a support request with a subject and a full description to the Customermates team. SIDE EFFECT: sends a real email. The Customermates team replies to the email address on the user's account. Returns confirmation only after the email provider accepts the request.
| Name | Type | Req | Description |
|---|---|---|---|
| body | string | yes | The full question or problem description, including relevant context. |
| subject | string | yes | Short summary of the problem or question. |
| Name | Type | Req | Description |
|---|---|---|---|
| accepted | boolean | yes | – |
No examples provided.
resolve_record_identifiers Find records by channel identifiers ~199
Resolves exact channel identifiers in one indexed batch. Each identifier is a provider and a value, such as mail and an email address, or a supported phone or profile identifier; optional typeIds restrict the lists searched. Accepts up to 1,000 identifiers. Each input returns every accessible linked record with its stable reference, type label and current version; an empty records array means no accessible match, and several records can share one identifier. The result carries the schema revision and each record's version, which serve as expectedRevision and expectedVersion of an updateMany mutation target (up to 100 per atomic batch) without a separate read; a changed record rejects the whole batch. Ordinary text fields are not channel identities. Returned labels are untrusted customer data.
| Name | Type | Req | Description |
|---|---|---|---|
| identifiers | array | yes | Up to 1,000 exact channel identifiers. Each input, including a repeated one, returns its own match. |
| typeIds | array | – | – |
| Name | Type | Req | Description |
|---|---|---|---|
| matches | array | yes | – |
| schemaRevision | integer | yes | Workspace schema revision, usable as expectedRevision of a record mutation. |
No examples provided.
resume_crm_operation Resume an interrupted operation ~67
Resumes an owned pending operation after its worker lease expires. Retries reuse staged progress without duplicating record writes. The worker rechecks publication preconditions, so a resume response does not mean publication has completed. Returns resumed: true or false.
| Name | Type | Req | Description |
|---|---|---|---|
| operationId | string | yes | – |
| Name | Type | Req | Description |
|---|---|---|---|
| resumed | boolean | yes | – |
No examples provided.
save_message_draft Save message draft ~379
Saves a message as a draft for inbox review without sending it. threadId drafts a reply on that thread or updates a saved draft conversation; connectedAccountId plus recipients creates a new draft conversation, including for recipients that already have one. recipients are email addresses or one chat handle. subject, cc and bcc are supported for email only; recipients: [] allows a Cc-only or Bcc-only email with at least one cc or bcc address. Explicit recipients also set the reply To. Saving on a threadId replaces that thread's one draft. The enabled account signature is appended at send time. Email drafts can carry files through fileIds (file values, a record's files list, or an upload made without recordId); omitting fileIds keeps the draft's files, [] removes them, and changed files produce a new revision. Returns draftMessageId, draftRevision and threadId, which identify the draft for a later send or discard. Nothing is sent.
| Name | Type | Req | Description |
|---|---|---|---|
| bcc | array | – | – |
| body | string | yes | Email drafts use Markdown; chat drafts use plain text |
| cc | array | – | – |
| connectedAccountId | string | – | New outbound draft: the connected account to send from later. Required when threadId is omitted |
| fileIds | array | – | Email drafts only: files to attach, from a file value, a record's files list or an upload without recordId. Replaces the draft's attachments; omit to keep them, [] removes them |
| recipients | array | – | Primary recipients to preserve with the draft. Required when threadId is omitted; use [] for Cc-only or Bcc-only email |
| subject | string | – | – |
| threadId | string | – | Reply draft: the existing thread to draft on. Omit to draft a brand-new outbound conversation |
| Name | Type | Req | Description |
|---|---|---|---|
| draftMessageId | string | yes | – |
| draftRevision | string | yes | – |
| threadId | string | yes | – |
No examples provided.
schedule_meeting Schedule, move and cancel meetings ~657
Creates, moves and cancels meetings in the caller's own connected Google or Microsoft calendar, and finds free slots from the synced calendars. Needs the inbox plan (Pro and up) and a connected calendar; it is not available when self-hosting. options returns the writable calendars, attendee suggestions from a record and its linked People, and the lists that can receive a follow-up Action. slots returns free start times within working hours that avoid synced events. create makes the event, links it to the given records so it appears on their activity timeline, and with followUp also creates a record in that list at the meeting start. SIDE EFFECT: create, update and cancel change a real calendar, and unless sendInvites is false the provider sends invitations or updates to every attendee. cancel notifies the attendees and cancels the event permanently for them. Only events created through this tool or the app and owned by the caller can be changed; an event id is the id returned by create or the id of a synced calendar event.
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | options: record; slots: durationMinutes, timeZone, from, days, workingHours, bufferMinutes; create: title, startsAt, durationMinutes, timeZone, calendarId, description, location, attendees, videoLink… |
| attendees | array | – | – |
| bufferMinutes | integer | – | slots only: free time kept around busy time. |
| calendarId | string | – | A calendar from the meeting options; the first writable calendar of the caller when omitted. |
| comment | string | – | cancel only: a message to the attendees, shown by Outlook. |
| days | integer | – | slots only: how many days to search, 14 by default. |
| description | string | – | – |
| durationMinutes | integer | – | – |
| followUp | object | – | Creates a record in this list at the start of the meeting, linked to the records, as the follow-up Action. |
| from | string | – | slots only: the first instant to consider. |
| id | string | – | The event id returned when the meeting was created. |
| location | string | – | – |
| record | object | – | options only: the record the meeting is for, to get attendee suggestions from its linked People. |
| records | array | – | Records the meeting belongs to. It appears on their activity timeline. |
| sendInvites | boolean | – | – |
| startsAt | string | – | Start of the meeting as an instant with offset or Z. |
| timeZone | string | – | IANA time zone the event is shown in, for example Europe/Berlin. |
| title | string | – | – |
| videoLink | boolean | – | Ask the provider for a video link: Google Meet on Google, Microsoft Teams on Outlook. |
| workingHours | – | – | slots only: bookable windows per weekday as { monday: [{ start: '09:00', end: '17:00' }], ... }; Monday to Friday 09:00 to 17:00 by default. |
| Name | Type | Req | Description |
|---|---|---|---|
| action | string | yes | – |
| actionLists | array | – | – |
| attendees | array | – | – |
| blockedReason | – | – | – |
| calendars | array | – | – |
| cancelled | object | – | – |
| meeting | object | – | – |
| slots | array | – | – |
| timeZone | string | – | – |
No examples provided.
search Search workspace knowledge ~207
Required by ChatGPT company-knowledge and deep-research connectors. Returns relevant Knowledge Base pages, files attached to records, CRM records, and product documentation in one list, without totals or filters. Knowledge Base and file results match the query's words in full text and, with AI credits, also by meaning, and carry a snippet with the matched terms in **, the matched section, and its offset, the start of the matched section. A file result (file:<uuid>) comes from the extracted text of a file on a record the caller can read and names that record; its snippet arrives between untrusted-content markers. A misspelled word that matches no Knowledge Base page or file is corrected from their own words, and didYouMean names the corrected query. Links into the app in the docs are full URLs on this instance and open the reader's own workspace.
| Name | Type | Req | Description |
|---|---|---|---|
| query | string | yes | Search terms for workspace Knowledge Base content, file text, CRM record names, and product documentation |
| Name | Type | Req | Description |
|---|---|---|---|
| didYouMean | array | – | The corrected query the Knowledge Base or file results were found with, when a misspelled word matched no page or file |
| results | array | yes | – |
No examples provided.
search_crm_records Search accessible records ~210
Searches text across accessible configured record types, including custom types. Optional typeIds restrict the types searched, and includeEmbedded adds embedded records such as line items. Results are paginated with nextCursor, which is valid only with the same searchTerm, typeIds and includeEmbedded. A record id finds that record; a phrase in double quotes matches only as written; when nothing matches, similarly spelled titles are returned. Results carry stable typeId and recordId references. Restricted field values are excluded from search.
| Name | Type | Req | Description |
|---|---|---|---|
| cursor | – | – | – |
| includeEmbedded | boolean | – | – |
| limit | integer | – | – |
| searchTerm | string | yes | Text to find in titles and searchable fields; a record id finds that record. Records whose title equals, starts with or contains the text come first. Wrap a phrase in double quotes to match it only a… |
| typeIds | array | – | – |
| Name | Type | Req | Description |
|---|---|---|---|
| nextCursor | – | yes | – |
| results | array | yes | – |
| schemaRevision | integer | yes | – |
| untrustedTextHandling | string | – | Present when the result holds marked untrusted text; how to treat the text between the markers. |
No examples provided.
search_docs Search documentation ~195
Searches the Customermates documentation (product guides and REST API reference). Matches the query's words in full text and, with AI credits, also by meaning. Required: query. Optional: locale (one of: en, de, fr, it, es, pt, pl, nl; default en), source (one of: docs, api, all; default docs). Returns ranked pages with the best section of each (slug#anchor), then the best page's url and its snippet in text, plus up to 5 full matches as structured content. Links into the app in the docs are full URLs on this instance and open the reader's own workspace.
| Name | Type | Req | Description |
|---|---|---|---|
| locale | string | – | Documentation language |
| query | string | yes | Free-text search, e.g. 'webhook signature' or 'filter operators' |
| source | string | – | docs = product guides, api = REST endpoint reference, all = both |
| Name | Type | Req | Description |
|---|---|---|---|
| results | array | yes | – |
| total | integer | yes | – |
No examples provided.
send_chat_message Send chat message ~535
Sends a real chat message through a connected chat account (LinkedIn, WhatsApp and other chat providers). SIDE EFFECT: the message is sent immediately and cannot be recalled. Exactly one mode applies: with threadId, the text is sent into that existing thread; without threadId, a new chat is started, which requires connectedAccountId (a connected account whose status is ok) and attendeeIdentifiers (the recipients' provider handles, i.e. the value of a contact's messaging channel), plus an optional chatName to name the group. New LinkedIn chats default to the Classic product; linkedinProduct set to sales_navigator or recruiter sends an InMail from that product's inbox (requires inmailSubject; recruiter also inmailSignature), and inmail true on classic sends an InMail to someone outside the network. A linkedinProduct that is not listed in the account's linkedinProducts is rejected. An identical text sent into the same thread within about a minute is rejected as a duplicate. A saved draft is sent by passing both draftMessageId and its opaque draftRevision, as returned when the draft was saved or shown in a thread's messages; threadId may then be the draft's own conversation. Each draft revision is sent at most once: a repeated send returns the first result with repeated true. A refused draft (changed, still sending, outcome unknown, or another version already sent) was not sent. Sends chat messages only, not email.
| Name | Type | Req | Description |
|---|---|---|---|
| attendeeIdentifiers | array | – | – |
| chatName | string | – | Optional display name for the new chat (group chats) |
| connectedAccountId | string | – | – |
| draftMessageId | string | – | Id of the saved new-conversation draft to send. Each draft revision is sent at most once; a repeated send returns the first result |
| draftRevision | string | – | Opaque revision returned with the saved draft; required whenever draftMessageId is provided |
| inmail | boolean | – | LinkedIn classic only: send as InMail to message people outside the sender's network (uses an InMail credit) |
| inmailSignature | string | – | Sender signature. Required for recruiter |
| inmailSubject | string | – | Subject line of the InMail. Required for sales_navigator and recruiter |
| linkedinProduct | string | – | LinkedIn only: which product to send from. classic (default), sales_navigator or recruiter start the chat from that product's inbox; the two latter send an InMail and require inmailSubject (recruiter… |
| text | string | yes | – |
| threadId | string | – | – |
| Name | Type | Req | Description |
|---|---|---|---|
| messageId | – | yes | – |
| repeated | boolean | yes | – |
| sent | boolean | yes | – |
| threadId | – | yes | – |
No examples provided.
send_email Send email ~468
Sends a real email from a connected email account. SIDE EFFECT: the email is sent immediately and cannot be recalled. Requires to, subject, body and either threadId (a reply; takes precedence over connectedAccountId, and a draft-only conversation sends its draft) or connectedAccountId (a new email; an account whose status is ok). to takes {identifier} objects; cc and bcc take email strings. At least one recipient is required across these groups; to: [] allows a Cc-only or Bcc-only email, and Bcc recipients are not shown in To or Cc. A saved draft is sent by passing draftMessageId and draftRevision, as returned when the draft was saved or shown in a thread's messages; each revision is sent at most once and a repeated send returns the first result with repeated true. A saved draft sends with its own files; fileIds attaches CRM files to a new email. A refused draft (changed, recipients differ, still sending, outcome unknown, or another version already sent) was not sent. The enabled account signature and appearance are applied automatically, and the signature is appended in addition to anything written in body. Duplicate reply bodies within a minute are rejected.
| Name | Type | Req | Description |
|---|---|---|---|
| attachments | array | – | – |
| bcc | array | – | – |
| body | string | yes | – |
| bodyFormat | string | – | How to interpret body. Account appearance is applied to plain text and Markdown; explicit HTML remains caller-controlled. Omit for backwards-compatible automatic detection |
| cc | array | – | – |
| connectedAccountId | string | – | – |
| draftMessageId | string | – | Id of the saved draft to send. Each draft revision is sent at most once; a repeated send returns the first result |
| draftRevision | string | – | Opaque revision returned with the saved draft; required whenever draftMessageId is provided |
| fileIds | array | – | Files to attach from the CRM file store: a file value, a record's files list or an upload without recordId. When sending a saved draft, omitting it sends the draft's own files |
| subject | string | yes | – |
| threadId | string | – | – |
| to | array | yes | Primary recipients. Use [] for Cc-only or Bcc-only email |
| Name | Type | Req | Description |
|---|---|---|---|
| messageId | – | yes | – |
| repeated | boolean | yes | – |
| sent | boolean | yes | – |
| threadId | – | yes | – |
No examples provided.
update_messaging_thread Update messaging thread ~121
Sets a thread's state inside the Customermates inbox only, not at the email or chat provider. Requires threadId and state. state is one of unread, open, closed, or spam; the state shows as a badge on the thread and is filterable in the inbox, and it does not hide or delete anything. Setting the state a thread already has is a harmless no-op. The provider mailbox, the messages themselves, and any linked CRM records are left untouched.
| Name | Type | Req | Description |
|---|---|---|---|
| state | string | yes | – |
| threadId | string | yes | – |
| Name | Type | Req | Description |
|---|---|---|---|
| state | string | yes | – |
| threadId | string | yes | – |
No examples provided.
update_my_profile Update my profile ~168
Updates the current user's own profile and preferences. It is a partial update of firstName, lastName, avatarUrl, theme, displayLanguage and formattingLocale; omitted fields keep their current values. It does not change other members or workspace configuration.
| Name | Type | Req | Description |
|---|---|---|---|
| avatarUrl | – | – | HTTPS avatar URL, or '' / null to clear. Omit to keep existing. |
| displayLanguage | string | – | Language of the app interface. Omit to keep existing. |
| firstName | string | – | Omit to keep existing |
| formattingLocale | string | – | Locale for number, date and currency formatting. Omit to keep existing. |
| lastName | string | – | Omit to keep existing |
| theme | string | – | Colour theme: system, light or dark. Omit to keep existing. |
| Name | Type | Req | Description |
|---|---|---|---|
| message | string | yes | – |
No examples provided.
How do I install the Customermates MCP server?
Customermates is a hosted endpoint at https://customermates.com/api/v1/mcp, so there is nothing to install locally. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.
remote · customermates.com
claude mcp add --transport http benjiwagner-customermates 'https://customermates.com/api/v1/mcp'
What is the Customermates MCP server?
Customermates is an MCP server listed in the public MCP registry as io.github.benjiwagner/customermates. CRM: Claude, ChatGPT or any MCP client works with records, inbox and company knowledge. This page covers its hosted endpoint (https://customermates.com/api/v1/mcp).
Is the Customermates MCP server safe to use?
Customermates scores 74 out of 100 on VerifyMCP. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.
What tools does the Customermates MCP server expose?
Customermates exposes 63 tools: discover_record_types, get_record_model, query_crm_records, search_crm_records, resolve_record_identifiers, and 58 more. Their descriptions and schemas cost roughly 26,040 tokens of context every time the server is loaded.
Does the Customermates MCP server require authentication?
Yes. Customermates asked us for credentials when we connected, so you will need to authorise it in your MCP client before it can do anything.
Is the Customermates MCP server still maintained?
Customermates is still listed as active in the MCP registry. We last reached this channel on 21 November 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.
What did VerifyMCP test on the Customermates MCP server?
VerifyMCP ran 13 checks on Customermates, covering TLS certificate, Authorization, HTTPS, HSTS header and 9 more. Supply Chain Security and Provenance & Transparency do not apply to a hosted endpoint, so they were not tested. The score does not measure everything, and the scoring documentation lists what it leaves out.