Carly MCP reference

148 tools: 64 read, 44 write, 40 sensitive write. Setup steps for each client are on the MCP server page.

Connection

Server URL: https://carlyassistant.com/mcp/
Transport: Streamable HTTP, stateless
Resource metadata: https://carlyassistant.com/.well-known/oauth-protected-resource/mcp
Authorization server: https://carlyassistant.com/.well-known/oauth-authorization-server

Authentication

OAuth 2.1 is the default. Clients use the authorization code flow with PKCE (S256) and can register themselves with dynamic client registration. An unauthenticated request gets 401 with a WWW-Authenticate header that points to the resource metadata. Access tokens are short-lived; clients renew them with the refresh token. An OAuth connection acts for the one Carly account that authorized it and can use every tool on this page.

API keys are the alternative for clients without OAuth. Create one in the Carly dashboard under Integrations, then API Keys, choosing only the scopes it needs. Send it as Authorization: Bearer <key>. Each tool requires the scope listed beside it below, and a call without that scope is refused. Keys expire after at most 365 days and can be revoked at any time.

Scopes

booking_pages:read: read booking pages and event types
booking_pages:write: create / update booking pages
bookings:read: list and view bookings
calendar:read: read connected calendars and events
calendar:write: create / update / delete calendar data
mail:read: search and read connected mailboxes
mail:write: draft, send, and organize mailbox messages
workflows:read: list workflows and run history
workflows:write: create / edit / run / pause / resume workflows
skills:read: list and view skills
skills:write: create / edit / delete skills
memory:read: read agent memory
memory:write: save / delete agent memory
tasks:read: list scheduled tasks, reminders, and connected provider tasks
tasks:write: create / update / delete scheduled tasks, reminders, and connected provider tasks
workspace:read: read to-dos, data tables, and files
workspace:write: add / edit / delete to-dos, data tables, and Drive / OneDrive files
contacts:read: list and view contacts
contacts:write: add / edit / delete contacts
settings:read: read preferences, signature, and aliases
settings:write: change preferences, signature, and aliases
account:read: read credit usage and connected integrations

Read, write, and sensitive tools

Every tool carries MCP annotations: title, readOnlyHint, destructiveHint, and openWorldHint.

Read: Looks things up and changes nothing. readOnlyHint: true.
Write: Creates or changes data inside the user's own account or connected apps, such as drafts, labels, to-dos, and CRM fields. Nobody else is contacted. destructiveHint: false.
Sensitive write: Sends something to other people, deletes or overwrites data, runs code, or starts an automation that keeps running. destructiveHint: true. Clients should ask the user to approve each call.

Side effects to know

Send tools deliver email at once. A sent email cannot be recalled.
Creating, moving, or deleting a calendar event can notify its attendees through the calendar provider.
Workflows and scheduled tasks keep running on their schedule until they are paused or archived.
run_python runs code in a sandbox and can call apps the user has connected to Carly.
Sharing tools and booking pages make content visible to the people or public link they name.

Plans and limits

Free accounts can use every read tool, organize mail and data, and create drafts.
Sending email requires a paid plan. Some workflow features do too. The error says so and links to the plan page.
Each account can make 60 tool calls a minute and 1,800 an hour.
Sending has a daily recipient cap. The error gives the count and when it resets.
Mail, calendar, and file tools need the matching account connected in Carly. If it isn't, the result says so and how to connect it.
No region restrictions.

Errors and status

A failed call returns an MCP result with isError: true. Its structured content looks like this:

{
  "success": false,
  "error": "Human-readable reason",
  "code": "tool_failed",
  "retryable": false,
  "effects_may_have_occurred": false,
  "retry_after_seconds": 30
}

retry_after_seconds appears only when a retry is safe. When effects_may_have_occurred is true, do not retry automatically: an email may already have been delivered. Invalid arguments return an error naming the argument. Going over the rate limit returns Rate limit exceeded — retry in Ns. Booking-page tools report failure as {"ok": false, "error": …, "code": …} in the result instead.

tool_failed: The action failed. The message says why.
uncertain_effects: The action may have partly happened. Check before retrying.
capacity_exhausted: The code sandbox is busy. Retry after the delay given.
invalid_code: Code passed to a code tool was rejected before it ran.
preparation_failed: The code sandbox could not be prepared.
execution_failed: Code ran and raised an error.
job_too_large: The code job exceeded its size limit.
job_timeout: The code job ran past its time limit.
output_too_large: The result was too large to return.
artifact_persistence_failed: The action ran but its output file could not be saved.

Tools

Updated 2026-10-05. Open a tool to see its inputs.

add_todo Write workspace:write Add one item to the user's to-do list.

title required: string The to-do text, e.g.
due: string Optional due date, ISO 8601 (YYYY-MM-DD).
priority: integer 1 (urgent) to 4 (none).
project: string The list this belongs to, e.g.
labels: string[] Short tags, e.g.
notes: string Longer detail or context that doesn't belong in the title.
parent_id: integer Make this a sub-task of that to-do (a row_id from list_todos).
on: string[] CRM records this to-do is about: people by email, or record ids (deals, companies).
repeat: string Make it repeat: 'every day', 'every weekday', 'every monday', 'every mon, wed', 'every 2 weeks', 'every month', 'every year'.
due_time: string Optional time of day it is due, 24-hour 'HH:MM'.
assignee: string In a team workspace: the teammate it is for, by name or email.
workspace: string The team workspace to act in, by name (see Workspaces).

aggregate_crm Read contacts:read Calculate grouped counts or numeric aggregates over CRM records.

group_by required: string Field key to group by (e.g., 'status', 'company', 'source').
count_only: boolean If True, just return counts.
limit: integer Max number of groups to return.
filter_field: string Optional field to filter on before grouping.
filter_value: string Value to filter by (case-insensitive contains).

append_onedrive_workbook_row Write workspace:write Append one row to a OneDrive Excel worksheet, optionally seeding a header row.

file_id required: string OneDrive item id of the .xlsx workbook (from list_onedrive_files).
sheet_name required: string Worksheet/tab name to append to.
values required: object[] One row to append — a list of cell values, left to right (e.g.
headers: string[] Column headers to write as the first row, used ONLY when the sheet is empty.
drive_id: string Drive id for a workbook in a SharePoint/team document library (use the drive_id from resolve_onedrive_share_link); omit for the user's own OneDrive.
onedrive_account_id: integer Native account_id from list_integrations(query='OneDrive').

append_to_google_doc Write workspace:write Append text to the end of an existing Google Doc, found by id or name.

drive_account_id: integer Native account_id from list_integrations(query='Google Drive').
document_id: string Google Doc ID to append to.
document_name: string Find the Google Doc by name instead of ID.
folder_id: string Drive folder ID to search in when finding by document_name.
folder_name: string Folder name to search in when finding by document_name.
text required: string Text to append to the end of the document.

archive_workflow Sensitive write workflows:write Archive one owned workflow and retain its run history.

workflow_id required: integer ID of the workflow to archive.

bulk_mail_action Sensitive write mail:write Preview and then apply one mail change across every message matching a scope and filters — the batch alternative to calling a one-message tool repeatedly.

operation required: "preview" | "apply" 'preview' first, then 'apply' with only the returned snapshot_id.
mailbox: string Native Gmail/Outlook mailbox from list_mail_accounts(native_only=true), such as gmail:22 or outlook:33.
snapshot_id: string Required for apply; returned by a preview.
scope: string Required for preview: inbox | folder | all_mail.
location: string Gmail label ID/full name or Outlook folder path.
filters: object Narrow the scope: from_address, to_address, subject_contains, query (plain keywords, not provider syntax), received_after / received_before (ISO-8601).
action: string Required for preview: label | file | categorize | trash.
destination: string Gmail label ID/full name, Outlook category, or Outlook folder path.

cancel_scheduled_task Sensitive write tasks:write Cancel one scheduled task by its exact identifier.

task_id required: integer ID of the task to cancel.
force: boolean Cancel even though the task's thread is still waiting on a reply.
reason: string Why the task is being cancelled, in one user-facing sentence (e.g.

check_calendar Read calendar:read Check events on queryable calendars (e.g., conference rooms).

start_time required: string Start of time range.
end_time required: string End of time range.
calendar_ids: string[] Specific queryable calendar IDs to check.

complete_todo Write workspace:write Mark a to-do done by its row_id (from list_todos).

row_id required: integer The to-do's row_id, from list_todos.
workspace: string The team workspace to act in, by name (see Workspaces).

connect_integration Read account:read Where the user connects an app to Carly: ``connect_url``, the Carly Integrations page.

toolkit required: string
auth_scheme: string Deprecated and ignored.

connect_mcp_server Write settings:write Connect an OAuth-enabled public MCP server such as Granola from its HTTPS server URL.

server_url required: string The provider's public HTTPS MCP server URL, for example https://mcp.granola.ai/mcp.

copy_google_doc Write workspace:write Copy a Google Doc template with its formatting into a new working document.

document_id required: string
drive_account_id: integer Native account_id from list_integrations(query='Google Drive').
title required: string Title for the new working copy.
folder_id: string Destination Drive folder ID; omitted keeps the provider's default copy location.

create_availability_poll Write calendar:write Create a group-scheduling tracker for a set of participants and return mutual free slots.

title required: string Title for the internal scheduling tracker.
dates required: string[] List of date strings (YYYY-MM-DD) that are candidate days
daily_start_hour: integer Earliest hour of the day (0-23).
daily_end_hour: integer Latest hour of the day (1-24).
timezone: string IANA timezone string
participant_emails: string[] Other participants to invite.

create_booking_page_tool Write booking_pages:write Create a public booking page others can use to book time on the owner's calendar.

payload required: object POST body for `/api/v1/booking-pages` + the MCP `create_booking_page_tool`.

create_calendar Write calendar:write Create a new calendar in one of the user's connected calendar accounts.

name required: string Name for the new calendar (e.g.
description: string Calendar description
timezone: string Calendar timezone (e.g.
color: string Calendar color.

create_column Write contacts:write Create one typed custom column in the user's CRM.

label required: string Display name for the column (e.g.
field_type: string Field type: 'text', 'date', 'select', 'textarea', or 'boolean'
options: string[] Options for 'select' type fields (e.g.
is_filterable: boolean Whether this column appears as a filter option

create_data_table Write workspace:write Create a typed data table for non-contact records such as deals, cases, or ledgers.

label required: string Display label for the table, e.g.
name: string Optional snake_case table name.
description: string Short purpose/meaning of this table.
columns: object[] Column specs: [{key, label, field_type, options?, is_filterable?, is_name_field?}].
rows_file: string Optional fh_ CSV/TSV/JSON handle to import while creating the table.
key_field: string Optional natural upsert key column, e.g.
if_exists: "error" | "ignore" What to do if a table with this name already exists.

create_drive_folder Write workspace:write Create a Google Drive folder, reusing an existing folder of the same name when present.

drive_account_id: integer Native account_id from list_integrations(query='Google Drive').
folder_name required: string Name of the folder to create.
parent_folder_id: string Parent folder ID.
parent_folder_name: string Parent folder name.
force_create: boolean By default, creation is blocked when similarly named folders already exist (they are returned in similar_folders so you can reuse one).

create_event Sensitive write calendar:write Create a calendar event with explicit times, attendees, location, recurrence, and conferencing options.

summary required: string
start_time required: string Event start time in YYYY-MM-DDTHH:MM format.
end_time required: string Event end time in YYYY-MM-DDTHH:MM format.
description: string
location: string
attendees: string[]
calendar_id: string Use 'default' to create on the calendar associated with the sender's email account.
timezone: string IANA timezone for this event (e.g., 'America/New_York', 'Europe/Paris').
start_timezone: string IANA timezone for start_time ONLY.
end_timezone: string IANA timezone for end_time ONLY.
recurrence: string[] Recurrence rule for recurring events in RRULE format (RFC 5545).
is_video_call: boolean Set to True for video calls.
show_as_free: boolean If True, marks the event as free/transparent so it will not block calendar availability.
is_all_day: boolean Set to True for all-day events (e.g., vacations, holidays, multi-day trips).

create_gmail_draft Write mail:write Create one unsent draft in the user's Gmail Drafts folder for them to review and send.

gmail_account_id: integer Connected Gmail account ID to use.
drive_account_id: integer Connected Drive account ID to use when attaching Drive files.
to required: string[] Recipient email addresses.
subject required: string Email subject line.
body required: string Email body (plain text).
cc: string[] CC recipients.
bcc: string[] BCC recipients.
gmail_attachments: object[] Legacy — use `files` with the handle_token from list_gmail_attachments instead.
drive_file_ids: string[] Legacy — use `files` with the handle_token from list_drive_files instead.
files: string[] File-handle tokens (fh_...) to attach — the one param for every file: Gmail attachments, Drive files, uploads, saved attachments, run_python outputs.

create_gmail_drafts Write mail:write Create one personalized unsent Gmail draft per recipient, filled from each recipient's fields.

gmail_account_id: integer Connected Gmail account ID to use.
recipients: object[] List of recipient objects.
crm_contact_ids: integer[] CRM contact ids to draft for (from search_crm _record_ids or query_crm contact_id).
subject required: string Email subject line.
body required: string Email body (plain text).
files: string[] File-handle tokens (fh_...) to attach — chat uploads, saved attachments, or run_python outputs.

create_google_doc Write workspace:write Create a Google Doc in Drive, optionally in a named folder and with initial content.

drive_account_id: integer Native account_id from list_integrations(query='Google Drive').
title required: string Title of the Google Doc to create.
folder_id: string Drive folder ID to create the doc in.
folder_name: string Folder name to create the doc in.
content: string Optional initial text content for the document.

create_microsoft_planner_bucket Write tasks:write Create a bucket in an accessible basic Microsoft Planner plan.

microsoft_planner_account_id: integer Connected Microsoft Planner account id.
plan_id required: string
name required: string
order_hint: string

create_microsoft_planner_plan Write tasks:write Create a basic Planner plan in a Microsoft 365 group the signed-in user already belongs to.

microsoft_planner_account_id: integer Connected Microsoft Planner account id.
title required: string
group_id required: string Microsoft 365 group id.

create_microsoft_planner_task Write tasks:write Create and optionally assign a task in a basic Microsoft Planner plan.

microsoft_planner_account_id: integer Connected Microsoft Planner account id.
plan_id required: string
title required: string
bucket_id: string
assignee_ids: string[] Microsoft Entra user ids to assign.
start_at: string ISO 8601 UTC date/time.
due_at: string ISO 8601 UTC date/time.
priority: integer
description: string

create_microsoft_todo_list Write tasks:write Create a list in the explicitly connected Microsoft To Do account.

microsoft_todo_account_id: integer Connected Microsoft To Do account id.
display_name required: string Name for the new Microsoft To Do list.

create_microsoft_todo_task Write tasks:write Create a Microsoft To Do task with optional dates, reminder, recurrence, and notes.

microsoft_todo_account_id: integer Connected Microsoft To Do account id.
task_list_id required: string Microsoft To Do task-list id from list_microsoft_todo_lists.
title required: string
body: string
importance: "low" | "normal" | "high"
start_at: string Local ISO date/time in time_zone.
due_at: string Local ISO date/time in time_zone.
reminder_at: string Local ISO date/time in time_zone.
time_zone: string
recurrence_pattern: object
recurrence_range: object

create_onedrive_folder Sensitive write workspace:write Create a OneDrive folder, reusing an existing folder of the same name when present.

onedrive_account_id: integer Native account_id from list_integrations(query='OneDrive').
folder_name required: string Name of the folder to create.
parent_folder_name: string Parent folder name.
share_with: string[] Optional email addresses to grant Edit access on the folder once it exists.
force_create: boolean By default, creation is blocked when similarly named folders already exist (they are returned in similar_folders so you can reuse one).

create_outlook_draft Write mail:write Create one unsent draft in the user's Outlook Drafts folder for them to review and send.

outlook_mail_account_id: integer Which Outlook Mail account to use (id from connected accounts).
to required: string[] Recipient email addresses.
subject required: string Email subject line.
body required: string Email body (plain text).
cc: string[] CC recipients.
bcc: string[] BCC recipients.
ses_attachment_ids: integer[] Legacy — use `files` with the attachment's fh_ token instead.
outlook_attachments: object[] Legacy — use `files` with the handle_token from list_outlook_attachments instead.
onedrive_file_ids: object[] Legacy — use `files` with the handle_token from list_onedrive_files / upload_to_onedrive instead.
attachments: object[] Raw attachment refs as [{url, filename, content_type?}] for a public download URL.
files: string[] File-handle tokens (fh_...) to attach — the one param for every file: Outlook attachments, OneDrive files, uploads, saved attachments, run_python outputs.

create_outlook_drafts Write mail:write Create one personalized unsent Outlook draft per recipient, filled from each recipient's fields.

outlook_mail_account_id: integer Which Outlook Mail account to use (id from connected accounts).
recipients: object[] List of recipient objects.
crm_contact_ids: integer[] CRM contact ids to draft for (from search_crm _record_ids or query_crm contact_id).
subject required: string Email subject line.
body required: string Email body (plain text).

create_workflow Sensitive write workflows:write Create a recurring job with fixed steps: a trigger, a step graph, and account bindings.

name required: string Short user-visible name for the workflow.
team_id: integer Optional company/team scope id.
graph: object OMIT this when template_id is set — the server instantiates the blueprint's graph directly (preferred: the graph never passes through your context, so it can't…
trigger_type: string 'manual', 'cron', 'webhook', 'booking', 'phone', 'composio', 'pipedream', 'calendar_event', 'calendar_reminder', 'email_received', 'http_poll', 'tabular', 'for…
trigger_spec: object For cron workflows, pass {'cron': '0 9 * * 1'}.
timezone: string IANA timezone for cron schedules and ${now.*}.
run_now: boolean When true, create a manual run immediately after saving.
status: string Initial status: active or paused.
template_id: string PREFERRED creation path: set this to a template/recipe id and OMIT graph — the server instantiates the blueprint itself and adopts its suggested trigger unless…
template_params: object Template parameters.

create_workflow_from_template Sensitive write workflows:write Create a workflow from a vetted server-side Carly template.

name required: string User-visible name for the workflow.
template_id required: string Exact template id returned by list_workflow_templates.
template_params: object Values for the template's blanks, as '<node_id>.<field>': value and 'inputs.<key>': value entries — get_workflow_blueprint lists every accepted key under bluep…
trigger_type: string Trigger type.
trigger_spec: object Account bindings or schedule fields required by the selected template trigger.
timezone: string Optional IANA timezone override.

delete_booking_page_tool Sensitive write booking_pages:write Deactivate a booking page so its link stops accepting new bookings.

event_type_id required: integer

delete_data_rows Sensitive write workspace:write Delete selected rows from one owned data table.

table required: string CRM table name.
filters required: object Exact-match filters.

delete_event Sensitive write calendar:write Delete one calendar event by its exact identifier, optionally notifying attendees.

event_id required: string
calendar_id: string
notify_attendees: boolean
allow_invitation_removal: boolean Set true only when the user authorized removing their copy of an event organized by someone else.

delete_memory Sensitive write memory:write Delete one saved memory document by its exact key.

key required: string Doc name to delete

delete_microsoft_planner_task Sensitive write tasks:write Permanently delete a Microsoft Planner task using its last known ETag.

microsoft_planner_account_id: integer Connected Microsoft Planner account id.
task_id required: string
etag: string Last known task ETag; Carly fetches it if omitted.

delete_microsoft_todo_list Sensitive write tasks:write Permanently delete a Microsoft To Do list by provider list id.

microsoft_todo_account_id: integer Connected Microsoft To Do account id.
task_list_id required: string Microsoft To Do task-list id from list_microsoft_todo_lists.

delete_microsoft_todo_task Sensitive write tasks:write Permanently delete one Microsoft To Do task by provider ids.

microsoft_todo_account_id: integer Connected Microsoft To Do account id.
task_list_id required: string Microsoft To Do task-list id from list_microsoft_todo_lists.
task_id required: string Microsoft To Do task id from list_microsoft_todo_tasks.

delete_todo Sensitive write workspace:write Remove a to-do entirely by row_id (from list_todos), along with any sub-tasks under it.

row_id required: integer The to-do's row_id, from list_todos.
workspace: string The team workspace to act in, by name (see Workspaces).

describe_workflow_trigger Read workflows:read Inspect the payload a Composio, http_poll, or email_received trigger delivers, so graph steps can reference real ${trigger.*} field paths instead of guessed ones.

trigger_type: string Trigger type to inspect: composio, http_poll, or email_received.
composio_slug: string Composio trigger slug to inspect when trigger_type='composio' (e.g.
trigger_spec: object Trigger config to inspect when trigger_type='http_poll', e.g.
include_real_samples: boolean If true, fetch up to sample_limit recent items from the upstream source and shape them as the trigger would deliver.
sample_limit: integer Max real samples to fetch when include_real_samples is true (1-5).
account_alias: string Selector (alias or ca_* ID) for the connected account the workflow will use.

download_onedrive_file Read workspace:read Download a non-Excel OneDrive file: text inline, binaries as a readable stored attachment.

file_id required: string OneDrive item id (from list_onedrive_files).
drive_id: string Drive id the file lives in.
onedrive_account_id: integer Native account_id from list_integrations(query='OneDrive').
max_inline_bytes: integer Max bytes to return inline as text.

draft_mail_reply Write mail:write Create a draft reply on an existing Gmail or Outlook thread without sending it.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.
body required: string Reply body.
reply_all: boolean
cc: string[] Additional Cc recipients.
bcc: string[] Bcc recipients.

edit_google_doc Sensitive write workspace:write Replace exact text in an existing document using its observed revision.

document_id required: string
drive_account_id: integer Native account_id from list_integrations(query='Google Drive').
revision_id required: string Revision returned by the latest read_google_doc.
tab_id required: string Exact tab ID returned by read_google_doc.
replacements required: object[]

edit_todo Write workspace:write Change an existing to-do by row_id (from list_todos): retitle it, set/clear its due date, change priority/project/labels/description, or reopen it (status='open').

row_id required: integer The to-do's row_id, from list_todos.
title: string New title.
due: string New due date (YYYY-MM-DD), or '' to clear it.
status: string One of the list's column keys (from list_todos `statuses`): 'open' reopens, 'done' finishes; a team may have more.
priority: integer 1 (urgent) to 4 (none).
project: string Move to this project, or '' to remove it from one.
labels: string[] Replaces the label set.
notes: string New description, or '' to clear it.
parent_id: integer Move under this to-do (a row_id), or 0 to make it top-level again.
repeat: string Repeat rule ('every monday', 'every 2 weeks', 'every month'), or '' to stop repeating.
due_time: string Time of day it is due, 24-hour 'HH:MM', or '' for any time.
assignee: string In a team workspace: the teammate it is for, by name or email, or '' to unassign.
workspace: string The team workspace to act in, by name (see Workspaces).

find_event Read calendar:read Search the user's selected calendars for events matching text and time filters.

query required: string
start_time: string
end_time: string
max_results: integer
calendar_id: string If omitted, searches ALL selected calendars across the user's connected accounts (returns a deduplicated union).
timezone: string

find_files Read workspace:read Find the user's stored files by name, label, or tag and return ownership-scoped handles.

query: string Case-insensitive substring matched against a file's name, label, or tags — e.g.
tag: string Optional exact tag to filter by (e.g.
limit: integer Max files to return (capped at 25).

find_free Read calendar:read Find available time slots across the user's selected calendars in a requested window.

start_time: string Start of time window.
end_time: string End of time window.
timezone: string
max_results: integer
min_slot_minutes: integer

flag_mail_message Write mail:write Flag or unflag one message — a star in Gmail, a follow-up flag in Outlook.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.
flagged: boolean True to flag/star the message, false to clear it.

get_booking Read bookings:read Get one booking by uid: who booked, when, on which booking page, and its current status.

uid required: string

get_booking_page Read booking_pages:read Get one booking page's full configuration by id: duration, availability windows, questions asked of attendees, and its public link.

event_type_id required: integer

get_contact_emails Read contacts:read Return email addresses for an explicit set of owned CRM record ids.

contact_email: string Email address of the contact to look up (single-contact form).
contact_emails: string[] Batch form — look up MANY contacts in ONE call, e.g.
limit: integer Max threads to return per contact (default 5)

get_crm_schema Read contacts:read List the user's CRM columns and their field types.

include_distributions: boolean Whether to include value distributions for each field (counts of each distinct value).

get_microsoft_planner_plan Read tasks:read Read a basic Planner plan and its buckets/tasks.

microsoft_planner_account_id: integer Connected Microsoft Planner account id.
plan_id required: string
include_tasks: boolean
include_buckets: boolean
limit: integer

get_user_preferences Read settings:read Read the user's current Carly preferences document.

No inputs.

get_workflow Read workflows:read Get one owned workflow's current configuration and recent run summaries.

workflow_id required: integer ID of the workflow to inspect.
compact: boolean When true (default), tool.run_python / tool.run_calendar_code source bodies in `graph` are replaced by a {lines, sha256} stub — the wiring, ids, and every othe…

get_workflow_blueprint Read workflows:read Get one vetted workflow template and its required account bindings or parameters.

template_id: string Template id returned by list_workflow_templates.
config: object ONLY for parameterized recipe templates (rows with parameterized=true).

get_workflow_runs Read workflows:read List recent runs for one owned workflow, with optional step details.

workflow_id required: integer ID of the workflow whose recent runs should be returned.
limit: integer Maximum number of recent runs to return.
include_steps: boolean Default false: returns run-level status, nodes_visited, terminal_node_id, and on failure failed_node_id + failed_step_error_message — enough to triage health w…
include_raw: boolean Only used with include_steps=true.

grep_memory Read memory:read Search or list the user's saved Carly memory documents by text pattern or date range.

pattern: string Text to search for (case-insensitive; matches doc content and key).
start_date: string List/search docs on/after this date (YYYY-MM-DD).
end_date: string List/search docs on/before this date (YYYY-MM-DD).
limit: integer Max docs to return.

label_mail_message Write mail:write Apply one Gmail label or Outlook category to a message, optionally creating it.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.
label required: string Gmail label name or Outlook category to apply.
create_missing: boolean Create the label/category if it does not exist.

list_booking_pages Read booking_pages:read List personal and authorized team booking pages with their titles, durations, links, and whether each is currently accepting bookings.

No inputs.

list_bookings Read bookings:read List meetings people have booked on the account's booking pages, filterable by status, booking page, and time window.

status: string
event_type_id: integer
start_time: string
end_time: string
limit: integer

list_calendar_names Read calendar:read List the user's connected calendar accounts and calendars with their exact identifiers.

include_unselected: boolean If True, also include calendars the user has connected but not enabled for Carly (selected=False).

list_calendars Read booking_pages:read List connected calendars as `provider::account_id::calendar_id` keys.

No inputs.

list_data_tables Read workspace:read List the user's data tables with compact schemas and row counts, without loading rows.

include_contacts: boolean Include the physical contacts table in the schema listing.

list_drive_files Read workspace:read List or search files and folders in a connected Google Drive account.

drive_account_id: integer Native account_id from list_integrations(query='Google Drive').
query: string Optional filename search term (uses Drive name contains).
folder_id: string Drive folder ID to search in.
folder_name: string Folder name to search in.
limit: integer Max files to return (1-100).

list_event_types Read booking_pages:read List the API key owner's own event types.

username: string

list_events Read calendar:read List events from the user's selected calendars in a requested time range.

start_time: string Start of date range.
end_time: string End of date range.
max_results: integer
query: string Text search query to filter events.
output_format: "compact" | "full" Default 'compact' (id, title, start/end, time_label, location).
include_free_events: boolean Include Free events for duplicate checks or event management.
event_id: string Exact provider event id from a prior compact list_events result.
calendar_id: string Optional calendar filter.

list_integrations Read account:read List apps Carly can connect and whether the current user connected them.

query: string
connected_only: boolean

list_mail_accounts Read mail:read List the user's connected Gmail and Outlook mailboxes and their exact mailbox keys.

native_only: boolean True to list only native Gmail/Outlook mailboxes supported by bulk_mail_action.

list_mail_attachments Read mail:read List the attachments on one selected Gmail or Outlook message, each with an fh_ handle_token.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.

list_mail_folders Read mail:read List the folders in one selected mailbox, with the exact names move_mail_message accepts.

mailbox required: string Exact mailbox value returned by list_mail_accounts, such as 'gmail:22', 'outlook:33', 'gmail:ca_connection', or 'zoho:ca_connection:provider_account_id'.

list_mail_labels Read mail:read List the Gmail labels or Outlook categories available in one selected mailbox.

mailbox required: string Exact mailbox value returned by list_mail_accounts, such as 'gmail:22', 'outlook:33', 'gmail:ca_connection', or 'zoho:ca_connection:provider_account_id'.

list_mcp_connections Read account:read List the user's connected MCP servers, their status, and the number of available tools.

include_inactive: boolean Include connections that are still authorizing, need reconnection, or failed.

list_microsoft_planner_plans Read tasks:read List basic Microsoft Planner plans shared with the connected work/school account.

microsoft_planner_account_id: integer Connected Microsoft Planner account id.
limit: integer

list_microsoft_todo_lists Read tasks:read List Microsoft To Do lists.

microsoft_todo_account_id: integer Connected Microsoft To Do account id.
limit: integer

list_microsoft_todo_tasks Read tasks:read List tasks in one Microsoft To Do list, returning Microsoft provider ids.

microsoft_todo_account_id: integer Connected Microsoft To Do account id.
task_list_id required: string Microsoft To Do task-list id from list_microsoft_todo_lists.
status: "all" | "open" | "completed"
limit: integer

list_onedrive_files Read workspace:read List or search files and folders in a connected OneDrive account.

onedrive_account_id: integer Native account_id from list_integrations(query='OneDrive').
query: string Search term to filter files by name.
folder_id: string OneDrive item id of the folder to search in.
folder_name: string Folder path under root, e.g.
limit: integer Max files to return (1-100).

list_scheduled_tasks Read tasks:read List active scheduled agent tasks, optionally including completed and cancelled history.

include_history: boolean Set to true when the user asks about past tasks — completed one-offs, tasks the user cancelled, or failed tasks — e.g.

list_schedules Read booking_pages:read List personal and editable team working-hours schedules: each one's weekly hours, date overrides, timezone, whether it is the default, and which booking pages use it.

No inputs.

list_skills Read skills:read List the user's saved skills, whether each is prose or code, and the arguments it takes.

include_disabled: boolean Also list skills the user has switched off.

list_slots Read booking_pages:read List available slots for an owned event type.

start_time required: string
end_time required: string
username: string
event_type_slug: string
event_type_id: integer
duration: integer

list_todos Read workspace:read List the user's to-dos (open by default; pass status='all' for everything, or project= to scope to one list).

status: string 'open' (default: every column that is not done), 'all', or one of the list's column keys (returned as `statuses`).
project: string Only this project's items.
assignee: string In a team workspace: only this teammate's, by name or email.
workspace: string The team workspace to act in, by name (see Workspaces).

list_workflow_templates Read workflows:read Browse Carly's vetted workflow template library by category, tag, or text query.

query: string What the workflow should DO, in your own words — e.g.

list_workflow_trigger_calendars Read workflows:read List exact connected-calendar keys accepted by calendar-triggered workflow templates.

No inputs.

list_workflow_trigger_forms Read workflows:read List the user's Carly forms with the numeric form_id a form_submitted workflow trigger needs, plus each form's field keys.

No inputs.

list_workflow_trigger_mailboxes Read workflows:read List exact connected-mailbox keys accepted by email-triggered workflow templates.

No inputs.

list_workflows Read workflows:read List the user's workflows and compact recent health information.

include_archived: boolean Set true to include archived workflows.

load_skill Read skills:read Read one saved skill's full instructions, argument schema, and last run error.

skill_name required: string The name of the skill to load instructions for

lookup_person Read contacts:read Resolve a person or relationship from the user's contacts to an email address.

query required: string Name, relationship, or alias to look up (e.g., 'dad', 'John', 'my dentist')

manage_crm_records Sensitive write contacts:write Query, create, update, upsert, link, unlink, annotate, aggregate, or delete records across People, Companies, Deals, Notes, and custom CRM objects.

action required: string One of: list, get, create, update, upsert, delete, link, unlink, add_note, kanban, aggregate
object: string 'people', 'companies', 'deals', 'meetings', 'notes', a custom object name, or id.
record_id: integer get/update/delete/link: the record id (a person's contact id is their record id)
email: string People shortcut: find/create the person by email instead of record_id
values: object create/update: {attribute key: value}.
links: object create/update/link: {relation attribute key: [record ids or emails]} e.g.
match: string upsert: the ONE attribute key that identifies an existing record, e.g.
search: string list: word-based search; 3+ words allow one missing word
filters: object list: {attribute key: value} exact filters (stage by label)
conditions: object[] list/kanban/aggregate: [{key, op, value}] AND-ed together — the operator filters 'over $10k' or 'contacted this week' need.
sort: string list: attribute key or last_interaction_at / updated_at / created_at / name
order: string list: asc | desc
limit: integer list: max records (≤200)
offset: integer list: pagination offset
group_by: string aggregate: attribute key to group by (e.g.
sum_key: string aggregate: numeric/currency attribute to sum (e.g.
title: string add_note: title
body: string add_note: the note text
on: object[] add_note: record ids or people emails the note is about
workspace: string The team workspace to act in, by name (see Workspaces).

manage_crm_schema Sensitive write contacts:write Configure People, Companies, Deals, Notes, and custom CRM objects: list or create objects; add, edit, hide, reorder, or delete typed fields; configure deal stages and select options; and create, update, list, or delete saved table and boar…

action required: string One of: list_objects, create_object, add_attribute, update_attribute, delete_attribute, hide_attributes, show_attributes, reorder_attributes, set_options, set_…
object: string Object to act on: 'people', 'companies', 'deals', 'meetings', 'notes', a custom object's name, or its id.
label: string create_object: plural label (e.g.
singular_label: string create_object: singular label (e.g.
attribute: string Attribute key, label or id for update/delete/set_options/set_stages
attributes: object[] create_object: list of attribute specs [{label, field_type, options?, is_name_field?}]; hide/show/reorder_attributes: list of attribute keys (reorder = the ful…
field_type: string add_attribute: text, textarea, number, currency, date, datetime, select, multiselect, boolean, status, email, phone, url, relation
options: string[] select/multiselect options (add_attribute or set_options)
stages: object[] status stages for set_stages / add_attribute(status): [{label, kind: open|won|lost, color?}] — the full ordered list; existing stages are matched by label (or…
target_object: string relation attributes: which object the relation points to ('people', 'companies', 'deals', custom name)
multiple: boolean relation attributes: can a record link to many targets
is_filterable: boolean Show this attribute as a filter in the table
is_hidden: boolean update_attribute: hide (true) / show (false) in the table
delete_data: boolean delete_attribute: also remove the stored values from every record
view: string update_view/delete_view: the view's name or id
name: string create_view/update_view: the view name (e.g.
filters: object[] create_view/update_view: [{key, op, value}] AND-ed.
sort: object create_view/update_view: {key, order: asc|desc}
kind: string create_view/update_view: 'table' or 'board' (board needs a status attribute)

move_event Sensitive write calendar:write Move one event from its current calendar to a selected destination calendar.

event_id required: string ID of the event to move
source_calendar_id required: string ID or name of the calendar the event is currently on
destination_calendar_id required: string ID or name of the calendar to move the event to

move_mail_message Write mail:write Move one Gmail or Outlook message to an explicit label or folder.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.
destination required: string Existing Gmail label or Outlook folder name/path (list_mail_folders has the exact names).
create_missing: boolean Create the destination when supported if it does not exist.
archive: boolean Gmail only: also remove the message from the inbox after applying the destination label.

pause_workflow Write workflows:write Pause one owned workflow so future triggers stop firing.

workflow_id required: integer Exact workflow id returned by list_workflows.

preview_workflow Read workflows:read Preview a saved workflow or template graph without saving or performing external writes.

workflow_id: integer Optional ID of a saved workflow to preview.
graph: object Workflow graph to preview.
trigger_type: string Trigger type for the preview: manual, cron, webhook, booking, phone, composio, pipedream, calendar_event, calendar_reminder, email_received, http_poll, tabular…
trigger_spec: object Trigger config used to build the preview payload, e.g.
timezone: string IANA timezone used to render ${now.*} for unsaved graph previews.
sample_event: object Optional trigger payload; used for manual/cron (ignored for data triggers).
sample_index: integer For composio previews, which fetched sample event to use when sample_event is not provided.

query_availability_poll Read calendar:read Read one tracker's status, mutually available slots, and per-participant responses.

session_token required: string The public token of the scheduling session to query

query_crm Read contacts:read Query CRM rows with explicit structured filters, sorting, and pagination.

sql required: string A single read-only SQL query (SQLite dialect) over the user's CRM.
export_filename: string Optional CSV filename (e.g.

query_data Read workspace:read Run one read-only SQL SELECT over the user's own named data tables in a sandbox.

sql required: string A single read-only SQLite SELECT/WITH query over the selected tables.
tables: string[] Agent-created data table names to load into the query sandbox.
include_contacts: boolean Also load the physical contacts table when needed.
export_filename: string Optional CSV filename (e.g.

read_attachment Read workspace:read Read text from an ownership-checked stored file or Gmail/Outlook message attachment.

attachment_id: integer | string Integer SesAttachment.id.
file_handle: string Durable fh_ file token returned by tools such as run_python.
message_id: string Outlook or Gmail message ID containing the attachment.
email_attachment_id: string Optional — target a specific attachment by ID (from list_outlook_attachments / list_gmail_attachments).
provider: string Optional mail provider hint ('gmail' or 'outlook') for workflow-triggered messages.
account_id: integer | string Optional provider-neutral mail account id for workflow-triggered messages.
mailbox_key: string Mailbox identifier as returned by list_mail_accounts / list_mail_attachments, e.g.
outlook_mail_account_id: integer Which Outlook Mail account to use.
gmail_account_id: integer Which Gmail account to use.
page_range: string For PDFs: a page range like '1-10', '3', or '5-end'.
max_chars: integer Optional cap on returned characters.
search: string Case-insensitive term search within this one file (not a mailbox search).

read_google_doc Read workspace:read Read document text, tabs and revision before editing.

document_id required: string
drive_account_id: integer Native account_id from list_integrations(query='Google Drive').

read_knowledge_file Read workspace:read Read one knowledge document sequentially by chunk range, for paging or search context.

file required: string The document's numeric id (preferred) or title, as listed in the knowledge base catalog.
start_chunk: integer Chunk index to start reading from (0-based).
max_chunks: integer How many consecutive chunks to read (capped at 20).

read_mail_message Read mail:read Read one message from a selected Gmail or Outlook mailbox by exact message id.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.
folder_id: string Required for Zoho Mail: the folder_id returned by search_mail for this message.
offset: integer Body character offset for continuing a truncated read.
limit: integer Body characters to return.

read_memory Read memory:read Read one saved Carly memory document by its exact key.

key required: string Doc name to read

read_onedrive_workbook Read workspace:read List the worksheets in a OneDrive Excel file, or read one worksheet's rows.

file_id required: string OneDrive item id of the .xlsx workbook (from list_onedrive_files).
sheet_name: string Worksheet/tab name to read.
range: string Optional A1-style range, e.g.
max_rows: integer Max rows to return (1-1000).
drive_id: string Drive id the file lives in.
onedrive_account_id: integer Native account_id from list_integrations(query='OneDrive').

resolve_onedrive_share_link Read workspace:read Resolve a pasted OneDrive or SharePoint sharing link to its exact file_id and drive_id.

share_url required: string A SharePoint or OneDrive sharing link the user pasted (e.g.
onedrive_account_id: integer Native account_id from list_integrations(query='OneDrive').

resume_workflow Sensitive write workflows:write Resume one paused owned workflow so its configured triggers can fire.

workflow_id required: integer Exact workflow id returned by list_workflows.

run_calendar_code Sensitive write calendar:write Run a sandboxed snippet against the user's calendars for bulk or conditional edits.

code required: string Python code to execute against the user's calendar accounts.

run_pandas Read workspace:read Analyze the user's tabular data with pandas in Carly's sandbox.

attachment_id: integer Integer SesAttachment.id for a spreadsheet.
file_handle: string A file-handle token (`fh_...`) for the spreadsheet or JSON data to query — works for any kind: SES attachments, Drive, OneDrive, Gmail attachments, Outlook att…
expression required: string Python code to run against the loaded DataFrame.
sheet_name: string For Excel files with multiple sheets, the sheet name to load.

run_python Sensitive write workspace:write Run a Python snippet in Carly's sandbox with the user's connected clients bound.

code required: string Python code to execute.
args: object Optional dict bound as `args` inside the sandbox.
input_attachment_ids: integer[] Legacy param — prefer `files` for new code.
files: string[] Optional file-handle tokens (`fh_...`) to stage into `input_dir/` inside the sandbox.
workspace: string Optional durable working-set name, shared with run_code_interpreter.
publish_to_composio: boolean When True, every output file is also pushed into Composio's S3 storage.

run_skill Sensitive write skills:write Run one already-saved skill by name with explicit arguments.

skill_name required: string The name of the skill to execute.
args: object Arguments dict for the skill.
files: object[] Optional list of file-handle tokens (`fh_...`) to stage into the skill's `input_dir/`.
input_attachment_ids: integer[] Compatibility path: inbound SES attachment IDs to stage into `input_dir/` (runtime='python' only).
ses_attachment_ids: integer[] Alias for input_attachment_ids (merged with it).

run_workflow_now Sensitive write workflows:write Run one owned workflow immediately with its real side effects, outside its schedule.

input_data: object Manual workflows only: values for ${input.*}, separate from trigger data.
workflow_id required: integer ID of the workflow to run immediately.
sample_event: object Optional trigger payload override for non-manual workflows.
confirmed_message_id: string email_received only.
mailbox_key: string email_received only.
folder_id: string email_received only.

save_memory Sensitive write memory:write Create or replace one named memory document for future conversations.

key required: string Doc name (e.g.
content required: string Full doc content (markdown).

save_skill Write skills:write Save a sandbox Python snippet as a named reusable skill.

skill_name required: string Name of the skill.
runtime required: string Execution runtime.
code required: string Python code to store as the skill body.
tool_schema: object JSON-Schema-ish dict describing the args this skill expects.
description required: string One-line description used for skill matching (same field as prose skills).
instructions: string Markdown playbook explaining when to use the skill and what the args mean.
verification_notes required: string How you verified this code.

schedule_task Sensitive write tasks:write Schedule a one-time, cron, or interval agent task with an explicit prompt and delivery settings.

name required: string Short name for the task (e.g.
prompt required: string What the agent should do when this task runs.
schedule_type required: string 'once', 'cron', or 'interval'.
schedule_spec: object Schedule details.
delay_minutes: integer For 'once' tasks: run this many minutes from now.
timezone: string IANA timezone for the schedule.
delivery_channel: string How the task's output reaches the user when the fire-time run doesn't send anything itself: 'email' (default) or 'sms' (the final reply is texted to the user's…
status: string Initial status: active or paused.

search_crm Read contacts:read Search the user's CRM records by names, emails, companies, and text fields.

query: string Free-text search across contact name, email, notes, and all custom CRM data fields.
filters: object Dict of field_key → value to filter by specific fields (e.g.
sort_by: string Field to sort by: 'interaction_count' (most emailed), 'last_interaction_at' (most recent), or 'updated_at' (default).
limit: integer Max records to return inline (capped at 50).

search_knowledge_base Read workspace:read Search the user's uploaded knowledge documents by meaning and keyword.

query required: string Natural-language question or keywords to search the knowledge base for — e.g.
file: string Optional: restrict the search to one document.
max_results: integer Max passages to return (capped at 12).

search_mail Read mail:read Search one selected Gmail or Outlook mailbox with provider-neutral structured filters.

mailbox required: string Exact mailbox value returned by list_mail_accounts, such as 'gmail:22', 'outlook:33', 'gmail:ca_connection', or 'zoho:ca_connection:provider_account_id'.
query: string Plain-text mailbox search query.
from_address: string Only messages from this email address.
to_address: string Only messages sent to this email address.
subject: string Subject text filter.
within_days: integer Only messages from the last N days (1-36500).
inbox_only: boolean Restrict results to the inbox.
unread_only: boolean Return only unread messages.
has_attachment: boolean Filter by whether a message has attachments.
max_results: integer Max messages to return (1-250).
page_token: string Pagination token returned by a prior identical search.
include_body: boolean Include each message's FULL body in the results.

send_email Sensitive write mail:write Send a new email from Carly to explicit recipients, with optional attachments and idempotency key.

to: object[] Recipient objects.
crm_contact_ids: integer[] CRM contact ids to send to (from search_crm _record_ids or query_crm contact_id).
subject: string Email subject — used for brand-new threads.
body required: string Email body.
cc: string[] CC recipients.
send_individually: boolean When true, sends a separate email per recipient (each rendered with that recipient's own fields) — use for outreach, prospecting, or any case where recipients…
start_new_thread: boolean When false: the reply stays in the current thread and recipients see its subject.
allow_new_recipients_in_thread: boolean When false, sends to external recipients not already on the active thread are blocked.
allow_merge_fallbacks: boolean When false, multi-recipient individualized external sends fail if any referenced merge var is missing, even if the placeholder has a fallback like {{first_name…
on_behalf_of: string Email address of the user on whose behalf this email is being sent.
attachments: object[] File attachments.
ses_attachment_ids: integer[] Legacy — use `files` with the attachment's fh_ token instead.
outlook_attachments: object[] Legacy — use `files` with the handle_token from list_outlook_attachments instead.
onedrive_file_ids: object[] Legacy — use `files` with the handle_token from list_onedrive_files instead.
files: string[] File-handle tokens (`fh_...`) to attach — the one param for every file: inbound attachments, Gmail/Outlook attachments, OneDrive/Drive files, uploads, and run_…

send_gmail Sensitive write mail:write Send a new email from the user's own Gmail account, so it appears sent by them.

gmail_account_id: integer Connected Gmail account ID to use.
drive_account_id: integer Connected Drive account ID to use when attaching Drive files.
to: object[] Recipient objects.
crm_contact_ids: integer[] CRM contact ids to send to (from search_crm _record_ids or query_crm contact_id).
subject required: string Email subject line.
body required: string Email body (plain text).
cc: string[] CC recipients.
bcc: string[] BCC recipients.
send_individually: boolean When true, sends a separate email per recipient (each rendered with that recipient's own fields) — use for outreach, prospecting, or any case where recipients…
gmail_attachments: object[] Legacy — use `files` with the handle_token from list_gmail_attachments instead.
drive_file_ids: string[] Legacy — use `files` with the handle_token from list_drive_files instead.
files: string[] File-handle tokens (fh_...) to attach — the one param for every file: Gmail attachments, Drive files, uploads, saved attachments, run_python outputs.

send_mail_reply Sensitive write mail:write Send a reply on an existing Gmail or Outlook thread.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.
body required: string Reply body.
reply_all: boolean
cc: string[] Additional Cc recipients.
bcc: string[] Bcc recipients.

send_outlook_mail Sensitive write mail:write Send a new email from the user's own Outlook account, so it appears sent by them.

outlook_mail_account_id: integer Which Outlook Mail account to use (id from connected accounts).
to: object[] Recipient objects.
crm_contact_ids: integer[] CRM contact ids to send to (from search_crm _record_ids or query_crm contact_id).
subject required: string Email subject line.
body required: string Email body (plain text).
cc: string[] CC recipients.
bcc: string[] BCC recipients.
send_individually: boolean When true, sends a separate email per recipient (each rendered with that recipient's own fields) — use for outreach, prospecting, or any case where recipients…
ses_attachment_ids: integer[] Legacy — use `files` with the attachment's fh_ token instead.
outlook_attachments: object[] Legacy — use `files` with the handle_token from list_outlook_attachments instead.
onedrive_file_ids: object[] Legacy — use `files` with the handle_token from list_onedrive_files / upload_to_onedrive instead.
attachments: object[] Raw attachment refs as [{url, filename, content_type?}] for a public download URL.
files: string[] File-handle tokens (fh_...) to attach — the one param for every file: Outlook attachments, OneDrive files, uploads, saved attachments, run_python outputs.

share_calendar Sensitive write calendar:write Grant an email address an explicit access role on a selected calendar.

calendar_id required: string ID or exact name of the calendar to share
email required: string Email address of the person to share with
role: string Permission level.

share_onedrive_item Sensitive write workspace:write Grant an explicit email address view or edit access to one OneDrive file or folder.

item_id required: string OneDrive item id (folder or file) to share.
recipients required: string[] Email addresses to grant access to.
role: string Permission level: 'write' (Edit) or 'read' (View).
require_sign_in: boolean If true, recipients must sign in.
send_invitation: boolean If true, OneDrive emails an invitation.
invitation_message: string Optional note included in the invitation email (only used when send_invitation=true).
onedrive_account_id: integer Native account_id from list_integrations(query='OneDrive').

test_workflow Read workflows:read Execute safe read steps for one workflow while replacing every external write with a preview.

workflow_id: integer Optional saved workflow id to test.
graph: object Unsaved workflow graph to test.
trigger_type: string Trigger type for unsaved graph tests: manual, cron, webhook, booking, phone, composio, pipedream, calendar_event, calendar_reminder, email_received, http_poll,…
trigger_spec: object Trigger config for unsaved graph tests.
timezone: string IANA timezone used to render ${now.*} for unsaved graph tests.
sample_event: object Optional trigger payload; accepted for manual/cron only (REFUSED for data triggers — use sample_message_id or sample_run_id to test those against real data).
sample_message_id: string Provider id of a REAL message in the watched mailbox to test on (email_received triggers).
sample_run_id: integer A past run id of this workflow (see get_workflow_runs) whose stored trigger payload and run inputs are replayed in their original namespaces without repeating…
sample_index: integer Fetched sample index to use when sample_event is omitted.
up_to_node_id: string Optional node id.
include_steps: boolean Default false: returns the dry-run verdict + node-by-node effect_chain + summary (status counts, failed steps with errors, and write previews showing exactly w…

trash_mail_message Sensitive write mail:write Move one message to Gmail Trash or Outlook Deleted Items; reversible from the provider's UI.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.

unlabel_mail_message Write mail:write Remove one Gmail label or Outlook category from a message.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.
label required: string Gmail label name or Outlook category to remove.

unsubscribe_from_sender Sensitive write mail:write Trigger a sender's RFC 8058 one-click List-Unsubscribe link from one of their messages.

mailbox: string Exact mailbox key from list_mail_accounts, or a connected native email address; never a bare account number.
message_id: string Legacy message ID; keep its owning mailbox selector.
message_ref: string Preferred account-bound reference returned by search.

update_availability_poll Write calendar:write Edit an existing tracker's dates, hours, timezone, title, or participants in place.

session_token required: string The public token of the scheduling session to update
title: string New tracker title
dates: string[] Replacement list of candidate days (YYYY-MM-DD).
daily_start_hour: integer New earliest hour of the day (0-23)
daily_end_hour: integer New latest hour of the day (1-24)
timezone: string New IANA timezone string
add_participant_emails: string[] Participant emails to add (skipped if already present)
remove_participant_emails: string[] Participant emails to remove from the tracker

update_booking_page_tool Write booking_pages:write Change a public booking page's title, duration, questions, limits, or which schedule it books against (`schedule_id`; null = the owner's default).

event_type_id required: integer
payload required: object PATCH body for `/api/v1/booking-pages/<id>` + the MCP `update_booking_page_tool`.

update_calendar Write calendar:write Update the name, description, or settings of an existing calendar.

calendar_id required: string ID or exact name of the calendar to update
name: string New calendar name
description: string New calendar description
color: string New calendar color (see create_calendar for valid values)

update_contact Write contacts:write Add or update facts and structured fields for one or more contacts.

email: string Email address of the contact (single-contact form).
update_request: string What to add/update about this contact in natural language, e.g.
updates: object[] Batch form — update MANY contacts in ONE call, e.g.

update_crm Write contacts:write Update explicit fields on selected owned CRM records.

contact_email: string Email of the contact.
record_id: integer ID of an existing contact to update.
fields: object Key-value pairs of CRM fields to set.
activity: object Optional timeline entry to APPEND to the contact's activity log: {'type': 'email_sent'|'email_received'|'booking_created'|'booking_cancelled'|'booking_reschedu…

update_data_rows Write workspace:write Update explicit fields on selected rows of one owned data table.

table required: string CRM table name.
filters required: object Exact-match filters.
set_fields required: object Fields to set on matching rows.

update_event Sensitive write calendar:write Update one owned calendar event by its exact identifier; omitted fields remain unchanged.

event_id required: string
summary: string
start_time: string New start time (YYYY-MM-DDTHH:MM).
end_time: string New end time (YYYY-MM-DDTHH:MM).
description: string Event description/body text.
description_mode: string How to apply the description field.
location: string
attendees: string[]
calendar_id: string
timezone: string IANA timezone for this update (e.g., 'Asia/Tokyo').
start_timezone: string IANA timezone for start_time ONLY.
end_timezone: string IANA timezone for end_time ONLY.
same_day_only: boolean
preserve_duration: boolean
reason: string
user_id: string
is_video_call: boolean Set to True for video calls.
show_as_free: boolean If provided, updates whether the event should appear as free (transparent) or busy (opaque).
color_id: string Google Calendar event color, as the colorId string '1'-'11': 1=lavender, 2=sage, 3=grape, 4=tangerine, 5=banana, 6=flamingo, 7=peacock, 8=graphite, 9=blueberry…
update_all_instances: boolean

update_microsoft_planner_task Write tasks:write Update, assign, move between buckets, or complete a Planner task using safe ETag concurrency.

microsoft_planner_account_id: integer Connected Microsoft Planner account id.
task_id required: string
etag: string Last known task ETag; Carly fetches it if omitted.
title: string
bucket_id: string
assignee_ids: string[] Complete replacement assignee list.
start_at: string
due_at: string
priority: integer
percent_complete: integer
description: string

update_microsoft_todo_task Write tasks:write Update, complete, or reopen a Microsoft To Do task in place using its provider ids.

microsoft_todo_account_id: integer Connected Microsoft To Do account id.
task_list_id required: string Microsoft To Do task-list id from list_microsoft_todo_lists.
task_id required: string Microsoft To Do task id from list_microsoft_todo_tasks.
title: string
body: string
importance: "low" | "normal" | "high"
status: "notStarted" | "inProgress" | "completed" | "waitingOnOthers" | "deferred"
start_at: string
due_at: string
reminder_at: string
time_zone: string
recurrence_pattern: object
recurrence_range: object
clear_start: boolean
clear_due: boolean
clear_reminder: boolean

update_participant_availability Write calendar:write Record one participant's available or unavailable times on an existing tracker.

session_token required: string The public token of the scheduling session
participant_email required: string Email of the participant to update
participant_name: string Display name for the participant
available_ranges: object[] List of available time ranges, each with 'start' and 'end' in ISO 8601 format
unavailable_ranges: object[] Optional list of unavailable time ranges, each with 'start' and 'end' in ISO 8601 format
timezone: string IANA timezone for the provided ranges.

update_schedule Write booking_pages:write Change a personal or authorized team schedule: its name, timezone, weekly hours (`availability`), or date overrides.

schedule_id required: integer
payload required: object PATCH body for `/api/schedules/<id>`.

update_scheduled_task Sensitive write tasks:write Update the prompt, name, schedule, timezone, or status of one scheduled task.

task_id required: integer ID of the task to update.
prompt: string New prompt to replace the task's stored instructions.
name: string New display name for the task.
schedule_spec: object New schedule details (same shape as schedule_task).
timezone: string New IANA timezone for the schedule (e.g.
delivery_channel: string Switch how the task's output is delivered: 'email' or 'sms' (text the user's registered phone).

update_skill Write skills:write Replace one saved skill's instructions with a complete new version.

skill_name required: string The name of the skill to update
new_instructions required: string The complete new instructions (replaces the old content entirely)
new_description: string Optional new one-line description.

update_user_preferences Write settings:write Apply one explicit natural-language change to the user's Carly preferences.

update_request required: string Natural language description of what preference to add, update, or remove

update_workflow Sensitive write workflows:write Update one owned workflow's name, trigger, schedule, timezone, status, or whole graph.

workflow_id required: integer ID of the workflow to update.
name: string New workflow name.
graph: object The whole new graph, not a patch.
trigger_type: string manual, cron, webhook, booking, phone, composio, pipedream, calendar_event, calendar_reminder, email_received, http_poll, tabular, form_submitted or crm_record.
trigger_spec: object Omit unless rescheduling: passing it replaces the live cron/trigger.
timezone: string IANA timezone.
confirm_trigger_change: boolean Confirm a deliberate replacement of an external trigger type.
status: string Optional new status: active or paused.
settings: object {key: value} for the graph's `inputs` (the user's 'Set up this workflow' fields), no graph needed.

update_workflow_step Sensitive write workflows:write Edit, insert, or remove one step of an owned workflow's graph without rewriting the rest.

workflow_id required: integer ID of the workflow whose step you want to patch.
node_id required: string ID of the node inside the active version's graph to patch.
code: string New source for a tool.run_python node: it reads rendered placeholders as `args` and assigns a `result` dict.
args: object Keys merged into a tool.run_python or tool.run_skill node's input.args (replace_args=true overwrites the dict, e.g.
instructions: string New instructions for an llm.call node, in full.
replace_args: boolean Replace input.args instead of merging into it.

upload_to_drive Sensitive write workspace:write Upload ownership-checked email attachments or file handles to Google Drive.

drive_account_id: integer Native account_id from list_integrations(query='Google Drive').
attachment_ids: integer[] Specific SesAttachment IDs (inbound email/chat attachments) to upload.
file_handle_tokens: string[] File-handle tokens (`fh_...`) to upload — for run_python outputs, Gmail/Outlook message attachments, and other files the agent has access to.
all: boolean When neither attachment_ids nor file_handle_tokens are given, upload all attachments on the current email.
folder_name: string Target Drive folder name.
naming_convention: string Naming template for uploaded files.
subfolder_by: string Automatically organize uploads into subfolders by 'sender', 'date', or 'subject'.
conflict_behavior: "replace" | "rename" | "fail" What to do if a file with the same target name already exists in the destination folder.

upload_to_onedrive Sensitive write workspace:write Upload ownership-checked email attachments or file handles to OneDrive.

onedrive_account_id: integer Native account_id from list_integrations(query='OneDrive').
attachment_ids: integer[] Specific SesAttachment IDs (inbound email/chat attachments) to upload.
file_handle_tokens: string[] File-handle tokens (`fh_...`) to upload — for run_python outputs and other files the agent has access to.
all: boolean When neither attachment_ids nor file_handle_tokens are given, upload all attachments on the current email.
folder_name: string Target OneDrive folder name or path (e.g.
naming_convention: string Naming template for uploaded files.
subfolder_by: string Automatically organize uploads into subfolders by 'sender', 'date', or 'subject'.
conflict_behavior: "replace" | "rename" | "fail" What to do if a file with the same target name already exists at the destination.

upsert_data_rows Write workspace:write Create or update rows in one owned data table, matched on its key field.

table required: string CRM table name, e.g.
rows: object[] Rows to create/update.
rows_file: string fh_ CSV/TSV/JSON handle to import directly.

upsert_rows Write contacts:write Create or update CRM records using explicit matching fields.

rows: object[] List of contact rows to create or update in ONE call.
rows_file: string fh_ file-handle token of a CSV or JSON file whose rows to upsert (e.g.

Support and security: support@calbotservice.com