Every tool the Hypertask MCP server exposes at https://mcp.hypertask.ai/mcp, with its parameters. Requests need a Bearer token in the Authorization header.
Source: Live MCP tools/list. Schema version 5cd847fdeb5b.
Does: List accessible boards, read their column manifests and playbooks, create or rename boards, and manage board settings and members.
Use when: Use to discover project_id or section_id before ticket changes, or to build a board from a brief using the create action.
Do not use when: Do not use to move individual tickets or change a column; use hypertask_tasks or hypertask_sections instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. A project is a board and a section is a column; manifest returns ordered columns and their semantic roles. Creating a board accepts nested tasks and section names, and archive changes visibility without deleting it.
Optional board description (HTML if product requires)
labels
object[]
No
Optional. Use for area/track/component/risk (e.g. Frontend, Security). Do not encode workflow state in labels;that belongs in sections.
sections
object[]
Yes
Delivery workflow columns left-to-right (array order): intake → committed → building → verify → done. Use domain-specific titles (Linear/Jira-style), not phase buckets like Discovery/Foundation unless the user asked for a phased roadmap. Avoid bland generic names as the default choice.
source_summary
string
No
Optional audit trail; not used for inference
tasks
object[]
No
Starter tasks for a realistic delivery board. Large/multi-doc sources may produce many tasks (often 100–250+);see create_board tool description. Each task: exactly one of section_index or section_title; for new work, bias placement to Triage/Scheduled; avoid parking most cards in In development/In review on a greenfield board. Shipped usually empty unless user said work is done. Each task SHOULD include priority and estimate unless user opted out. tasks[] is flat; use create_task + parent_task_id after board creation for explicit parent/child.
team_id
string | integer
Yes
Team/workspace id (numeric or UUID string from get_user_context.teams)
Does: Orient to the authenticated user, their accessible boards and account connections, list every enabled feature-kind flag and core product destinations with verified URLs where known, or update their own profile.
Use when: Use at the start of a session to discover accessible projects and permissions, or when the user requests a profile change.
Do not use when: Do not use for filtered board lists or task searches; use hypertask_projects or hypertask_search instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. The features action returns the complete catalog up to 300 entries; larger catalogs use pagination. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules.
Parameter
Type
Required
Description
action
hello | get | profile | connections | features
Yes
Operation to perform on this resource.
input
object | object | object | object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Call hypertask_user_context with action: "hello" for a guided start in MCP v2 (consolidated tools, on for Everyone). The legacy catalog exposes hypertask_hello. GET https://mcp.hypertask.ai/api/mcp/hello returns the same guide.
Field
What it contains
agent
The agent's id and displayName. It is null for a personal user token.
startBoard
The newest board across all boards the token can see, with id, title, and boardUrl. A missing title falls back to the short name. It is null when no board is available.
nextSteps
Three suggested steps for write agents, or two for read-only agents (no create step). No board means an empty list. Each has step, a CLI command, an MCP tool, and MCP arguments.
Each step uses the tool name and argument shape advertised by tools/list for the same credential. For the fictional board Product Launch (9142), the default consolidated steps for a write agent are:
Step
CLI command
MCP tool
MCP arguments
See the board
hypertask task list --project 9142
hypertask_tasks
{"action":"list","input":{"project_id":9142}}
Create a first task
hypertask task create --project 9142 --title "My first task"
hypertask_tasks
{"action":"create","input":{"project_id":9142,"title":"My first task"}}
Read the board rules
hypertask project playbook 9142
hypertask_projects
{"action":"playbook","input":{"project_id":9142}}
The legacy catalog instead uses hypertask_list_tasks, hypertask_create_task, and hypertask_get_board_playbook, with arguments at the top level (no action or input wrapper). Read-only agents receive only the list and playbook steps.
The tool description asks MCP clients to say who they are signed in as and offer these steps. For a write agent, for example: "Signed in as Claude Code (agent of Maria Lopez). Your board is Product Launch. Would you like to see the board, create a first task, or read the board rules?"
Does: Without a kind, a reply target or an @mention of a person (and no images or attachments), an agent comment is saved to the hidden ticket agent log, which people do not see. To post for people, pass kind: question for a decision by a person, qa for a QA verdict, or done for the closing summary. List comments and activity, add replies, and edit or delete permitted comments.
Use when: Use to discuss a ticket or inspect its conversation; list with include_activity also retrieves work history. For a person to decide, add a ticket comment with a line starting Question: and @mention that person using a real mention, not just their name. The mention notification appears in their Inbox; when Decisions is enabled for them and the notification is visible, the unanswered ticket appears in Inbox, split Decisions: https://app.hypertask.ai/inbox?split=Decisions. A later non-agent comment by that person on the ticket removes it from Decisions. Ordinary mentions are not Decisions. Structured decision records are separate; creating one alone does not create this Question: + @mention workflow. Use a comment, not a page, to request a decision.
Do not use when: Do not use for ticket descriptions or unpublished drafts; use hypertask_tasks or hypertask_drafts instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. text is HTML unless content_type=markdown; comment_id is required to edit or delete, and ticket identifiers locate list/add targets. Ownership and board membership still apply. board_id is an alias for project_id.
Parameter
Type
Required
Description
action
list | add | update | delete
Yes
Operation to perform on this resource.
input
object | object | object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Files to attach after the operation succeeds. Each file needs filename, content_type, and exactly one of data (base64 bytes; 3145728 decoded bytes total) or url (HTTP(S), for files up to 15 MB).
board_id
integer
No
Alias for project_id.
comment_id
integer
No
Comment ID. Required for update and delete. Obtain from get_comments_for_task.
content_type
html | markdown
No
Input format for text. Defaults to HTML.
has_attachments
boolean
No
Set true when files will be attached after creation so an agent post stays a comment instead of going to the agent log
id
integer
No
Existing page identifier, an alias of page_id.
inline_images
boolean
No
The inline images value for this operation.
kind
question | qa | done
No
Without a kind, a reply target or an @mention of a person (and no images or attachments), an agent comment is saved to the hidden ticket agent log, which people do not see. To post for people, pass kind: question for a decision by a person, qa for a QA verdict, or done for the closing summary.
mentions
object[]
No
Required when text contains @mentions. Resolve via list_project_members, then include { user_id, display_name } for each.
project_id
integer
No
Board id discovered with projects list or user_context; usually numeric, while time reports accept its string form.
reply_to_comment_id
integer
No
For agent replies, the comment ID that invoked the agent. This routes the answer to the requester's Important inbox.
reply_to_invocation_id
integer
No
For agent replies, the agent Mentioned notification ID that invoked the agent. Use this when the mention was in the task description, which has no comment ID. Routes the answer to the requester's Important inbox.
task_id
integer
No
Numeric ticket id; get accepts an array, while other actions usually accept a single id.
text
string
No
Comment text for add/update. HTML or structural markdown; content_type can explicitly select either format.
ticket_number
string
No
Stable full ticket key such as HTPR-1234; tasks get accepts an array of keys. task_number is accepted as an alias.
unique_index
integer
No
Ticket number within a board; requires project_id. Prefer ticket_number when known.
Does: Read, create, update, move, assign and relate tickets, including context, trees and description versions.
Use when: Use when working with a known ticket or a filtered board queue; list finds overdue work and context includes comments and recent activity.
Do not use when: Do not use for fuzzy text discovery or editing comments; use hypertask_search or hypertask_comments instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Prefer ticket_number (for example HTPR-1234); get accepts one identifier or a list, while context requires task_id and project_id. Normal, Archive and Deleted describe storage state, not workflow columns; section names describe workflow. board_id is an alias for project_id.
Parameter
Type
Required
Description
action
list | get | context | tree | next | versions | restore | link | links | unlink | create | update | move | waiting_on | assign
Yes
Operation to perform on this resource.
input
object | object | any | any | object | any | any | object | object | object | object | object | any | any | object | object | object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Task assignments use numeric human user ids; saved view filters also accept managed agent UUIDs. These are not display names.
attachments
object[]
No
Files to attach after the operation succeeds. Each file needs filename, content_type, and exactly one of data (base64 bytes; 3145728 decoded bytes total) or url (HTTP(S), for files up to 15 MB).
board_id
integer
No
Alias for project_id.
content_type
html | markdown
No
Input format for description. Defaults to HTML.
description
string
No
Task description. HTML or structural markdown; content_type can explicitly select either format.
due_date
string
No
Due date in ISO 8601 format (e.g. "2026-03-10" or "2026-03-10T00:00:00Z").
Label IDs to set on the task. Get available labels from get_user_context or list_projects (each project includes labels). Replaces existing labels; empty array removes all.
parent_task_id
number
No
The parent task ID to create the task as a sub-task of. If not provided, the task will be created as a top-level task.
Task assignments use numeric human user ids; saved view filters also accept managed agent UUIDs. These are not display names.
attachments
object[]
No
Files to attach after the operation succeeds. Each file needs filename, content_type, and exactly one of data (base64 bytes; 3145728 decoded bytes total) or url (HTTP(S), for files up to 15 MB).
board_id
integer
No
Alias for project_id.
content_type
html | markdown
No
Input format for description. Defaults to HTML.
description
string
No
Task description. HTML or structural markdown; content_type can explicitly select either format.
due_date
string,null
No
Due date in ISO 8601 format (e.g. "2026-03-10"). Pass null to clear/remove the due date.
estimate
integer
No
Relative ticket size: 0 means no size; 2 through 6 represent XS through XL.
id
integer
No
Existing page identifier, an alias of page_id.
inline_images
boolean
No
The inline images value for this operation.
labels
string,number[]
No
Label IDs to set on the task. Get available labels from get_user_context or list_projects (each project includes labels). Replaces existing labels; empty array removes all.
parent_task_id
number
No
The parent task ID to update the task as a sub-task of.
priority
integer
No
Ticket priority index or supported priority name; 0=None, 1=Urgent, 2=High, 3=Medium, 4=Low.
project_id
integer
No
Board id discovered with projects list or user_context; usually numeric, while time reports accept its string form.
section_id
integer
No
Numeric workflow column id; discover it with projects manifest.
status
Normal | Archive | Deleted
No
Storage state for tasks or projects (Normal, Archive, Deleted), or decision state where supported; not a workflow column.
task_id
integer
No
Numeric ticket id; get accepts an array, while other actions usually accept a single id.
ticket_number
string
No
Stable full ticket key such as HTPR-1234; tasks get accepts an array of keys. task_number is accepted as an alias.
title
string
No
Human-readable title of the ticket, board, column, page or view.
unique_index
integer
No
Ticket number within a board; requires project_id. Prefer ticket_number when known.
Does: Create, list, update, publish or delete unpublished ticket comment or description drafts.
Use when: Use to prepare text for review before it becomes a visible ticket update.
Do not use when: Do not use to post immediately; use hypertask_comments add or hypertask_tasks update instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules.
Parameter
Type
Required
Description
action
create | list | update | publish | delete
Yes
Operation to perform on this resource.
input
object | object | object | object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: Create, read, update, list or search ticket pages, and inspect, restore or archive their versions.
Use when: Use for longer documents attached to tickets; get with format retrieves the current page content.
Do not use when: Do not use for conversation replies or ticket descriptions; use hypertask_comments or hypertask_tasks instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. page_id or id identifies a page; update supports if_version for optimistic concurrency. Concise reads summarize pages; request detailed for full content.
Parameter
Type
Required
Description
action
create | get | update | list | search | versions | restore | archive
Does: List, read, create, update or delete board columns, including their completion and automatic-assignment settings.
Use when: Use when the board structure must change or when you need the tasks in one column.
Do not use when: Do not use to move a ticket between columns; use hypertask_tasks update with section_id instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules. role filters list or checks get by the semantic role the manifest publishes (intake, backlog, ready, in-progress, ai-review, human-review, done, other); it follows the column title and cannot be set.
Parameter
Type
Required
Description
action
list | get | create | update | delete
Yes
Operation to perform on this resource.
input
object | object | object | object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
For get: include tasks in the column. Default true.
is_done
boolean,null
No
For update: tickets in this column are finished. Reports count them as completed, Inbox Zero clears them, and agents stop treating them as blocking. null clears the flag and falls back to guessing from the column name.
limit
integer
No
For get: max tasks to return. Default 50.
move_after_section_id
integer
No
For update: move column to appear after this section ID.
offset
integer
No
For get: pagination offset.
project_id
integer
No
Project/board ID. Required for list, get, create, update, delete.
query
string
No
Free-text search over the list. Same name on every list tool.
For list: return only columns with this semantic role. For get: fail unless the column has this role. The manifest publishes these roles and they follow the title, so they cannot be set.
section_id
integer
No
Section/column ID. Required for get, update, delete. Use list to find IDs.
sort
string
No
Sort as field or field:asc|desc, for example updatedAt:desc.
title
string
No
Column title. Required for create. Optional for update (rename).
For get: include tasks in the column. Default true.
is_done
boolean,null
No
For update: tickets in this column are finished. Reports count them as completed, Inbox Zero clears them, and agents stop treating them as blocking. null clears the flag and falls back to guessing from the column name.
limit
integer
No
For get: max tasks to return. Default 50.
move_after_section_id
integer
No
For update: move column to appear after this section ID.
offset
integer
No
For get: pagination offset.
project_id
integer
No
Project/board ID. Required for list, get, create, update, delete.
query
string
No
Free-text search over the list. Same name on every list tool.
For list: return only columns with this semantic role. For get: fail unless the column has this role. The manifest publishes these roles and they follow the title, so they cannot be set.
section_id
integer
No
Section/column ID. Required for get, update, delete. Use list to find IDs.
sort
string
No
Sort as field or field:asc|desc, for example updatedAt:desc.
title
string
No
Column title. Required for create. Optional for update (rename).
For get: include tasks in the column. Default true.
is_done
boolean,null
No
For update: tickets in this column are finished. Reports count them as completed, Inbox Zero clears them, and agents stop treating them as blocking. null clears the flag and falls back to guessing from the column name.
limit
integer
No
For get: max tasks to return. Default 50.
move_after_section_id
integer
No
For update: move column to appear after this section ID.
offset
integer
No
For get: pagination offset.
project_id
integer
No
Project/board ID. Required for list, get, create, update, delete.
query
string
No
Free-text search over the list. Same name on every list tool.
For list: return only columns with this semantic role. For get: fail unless the column has this role. The manifest publishes these roles and they follow the title, so they cannot be set.
section_id
integer
No
Section/column ID. Required for get, update, delete. Use list to find IDs.
sort
string
No
Sort as field or field:asc|desc, for example updatedAt:desc.
title
string
No
Column title. Required for create. Optional for update (rename).
For get: include tasks in the column. Default true.
is_done
boolean,null
No
For update: tickets in this column are finished. Reports count them as completed, Inbox Zero clears them, and agents stop treating them as blocking. null clears the flag and falls back to guessing from the column name.
limit
integer
No
For get: max tasks to return. Default 50.
move_after_section_id
integer
No
For update: move column to appear after this section ID.
offset
integer
No
For get: pagination offset.
project_id
integer
No
Project/board ID. Required for list, get, create, update, delete.
query
string
No
Free-text search over the list. Same name on every list tool.
For list: return only columns with this semantic role. For get: fail unless the column has this role. The manifest publishes these roles and they follow the title, so they cannot be set.
section_id
integer
No
Section/column ID. Required for get, update, delete. Use list to find IDs.
sort
string
No
Sort as field or field:asc|desc, for example updatedAt:desc.
title
string
No
Column title. Required for create. Optional for update (rename).
For get: include tasks in the column. Default true.
is_done
boolean,null
No
For update: tickets in this column are finished. Reports count them as completed, Inbox Zero clears them, and agents stop treating them as blocking. null clears the flag and falls back to guessing from the column name.
limit
integer
No
For get: max tasks to return. Default 50.
move_after_section_id
integer
No
For update: move column to appear after this section ID.
offset
integer
No
For get: pagination offset.
project_id
integer
No
Project/board ID. Required for list, get, create, update, delete.
query
string
No
Free-text search over the list. Same name on every list tool.
For list: return only columns with this semantic role. For get: fail unless the column has this role. The manifest publishes these roles and they follow the title, so they cannot be set.
section_id
integer
No
Section/column ID. Required for get, update, delete. Use list to find IDs.
sort
string
No
Sort as field or field:asc|desc, for example updatedAt:desc.
title
string
No
Column title. Required for create. Optional for update (rename).
Does: List, read, create, update, delete or switch saved board views and their filters or sorting.
Use when: Use to persist a board presentation or choose the user’s active view.
Do not use when: Do not use for a one-time task search; use hypertask_tasks list or hypertask_search instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules.
Ordered multi-field sorting configuration for a saved view.
subtask_setting
None | Parent | Flattened | Card | Flattened_Card
No
None hides subtasks and their count; Parent shows parent tasks with a subtask count; Flattened shows subtasks as rows; Card shows subtasks on parent cards; Flattened_Card does both.
title
string
Yes
Human-readable title of the ticket, board, column, page or view.
visibility
Public | Private
No
Whether a saved view is shared (Public) or private to its owner (Private).
visible_section_ids
integer[]
No
Numeric column ids to display in the saved board view.
Ordered multi-field sorting configuration for a saved view.
subtask_setting
None | Parent | Flattened | Card | Flattened_Card
No
None hides subtasks and their count; Parent shows parent tasks with a subtask count; Flattened shows subtasks as rows; Card shows subtasks on parent cards; Flattened_Card does both.
title
string
No
Human-readable title of the ticket, board, column, page or view.
view_id
string
Yes
Saved board view UUID discovered by views list.
visibility
Public | Private
No
Whether a saved view is shared (Public) or private to its owner (Private).
visible_section_ids
integer[]
No
Numeric column ids to display in the saved board view.
Does: List, read, create, update, delete or import reusable agent skills scoped to the account or a board.
Use when: Use to inspect or maintain an agent instruction document; import can preview selected remote skills with dry_run.
Do not use when: Do not use for board playbooks or ticket pages; use hypertask_projects or hypertask_pages instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules.
Does: Read an inbox, archive or restore notification items, or send a ticket to a user inbox.
Use when: Use to triage notifications or explicitly bring an existing ticket to a person’s attention.
Do not use when: Do not use to archive tickets or boards; notification_ids identify inbox items, not task ids.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules.
Parameter
Type
Required
Description
action
list | archive | unarchive | move_task
Yes
Operation to perform on this resource.
input
object | object | object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: Start, stop, pause or resume ticket timers, log minutes, inspect running timers and report time entries.
Use when: Use when the user asks to track work or summarize recorded time in a date range.
Do not use when: Do not use for estimates or due dates; use hypertask_tasks update instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. task_identifier accepts a numeric task id or ticket number; user_id accepts a numeric user id or me. log requires minutes; from/to select dates and timezone_offset_minutes controls date interpretation.
Does: List owned agent identities, inspect live team presence, or create, revoke, archive or permanently delete an owned agent.
Use when: Use for agent identity administration or to see which agents are active on a team.
Do not use when: Do not use for ticket assignees or webhook delivery; use hypertask_tasks assign or hypertask_webhooks instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Human or management credentials are required for administration; presence requires team_id (a positive integer number or non-empty UUID string). Creating returns a credential once, and delete is irreversible.
Does: Upload files to a ticket or one of its comments.
Use when: Use when an existing ticket or comment needs a file; create and update actions can also accept inline attachments.
Do not use when: Do not use to fetch arbitrary URLs or upload without a known target ticket.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. files contains upload objects with filename, content_type and exactly one of data (base64) or url, subject to existing request size limits. Partial upload failures include retry guidance; do not repeat a successful mutation.
Parameter
Type
Required
Description
action
attach
Yes
Operation to perform on this resource.
input
object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: List custom field definitions on a board or set one custom field value on a ticket.
Use when: Use when a ticket needs board-specific typed metadata rather than its standard priority or due date.
Do not use when: Do not use for ordinary ticket fields; use hypertask_tasks update instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules.
Parameter
Type
Required
Description
action
list | set
Yes
Operation to perform on this resource.
input
object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: Create, list, read, answer or cancel structured decision requests attached to tickets.
Use when: Use when a ticket needs a recorded question with explicit options and a selected answer.
Do not use when: Do not use for ordinary discussion; use hypertask_comments instead. For a person to decide, add a ticket comment with a line starting Question: and @mention that person using a real mention, not just their name. The mention notification appears in their Inbox; when Decisions is enabled for them and the notification is visible, the unanswered ticket appears in Inbox, split Decisions: https://app.hypertask.ai/inbox?split=Decisions. A later non-agent comment by that person on the ticket removes it from Decisions. Ordinary mentions are not Decisions. Structured decision records are separate; creating one alone does not create this Question: + @mention workflow. Use a comment, not a page, to request a decision.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules.
Parameter
Type
Required
Description
action
create | list | get | answer | cancel
Yes
Operation to perform on this resource.
input
object | object | object | object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: List the labels on a board or create a new board label.
Use when: Use to discover valid label names before filtering or labeling tickets.
Do not use when: Do not use to attach labels to a ticket; use hypertask_tasks create or update instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules.
Parameter
Type
Required
Description
action
list | create
Yes
Operation to perform on this resource.
input
object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: Files anonymous, deduplicated feedback about Hypertask and returns the ticket receipt.
Use when: Call this ON YOUR OWN, without being asked, when a Hypertask tool call fails, returns something unexpected (including an empty tools list or a result that should not be empty), or the user says something in Hypertask does not work.
Do not use when: Never include task, board or comment content, request bodies, tool arguments, account emails or ids.
Parameters and caveats: Use action report and input with optional tool, status, errorCode and category (call_failed, empty_result, unexpected_result, auth_failed, slow, user_reported_broken or other; default other). Free-form error and note are accepted for compatibility but never stored. Structured fields must not contain credentials. Surface is mcp and the client comes from MCP clientInfo when available. Repeated errors bump an open ticket. Account hourly limits are 10 new reports and 30 bumps.
Parameter
Type
Required
Description
action
report
Yes
Operation to perform on this resource. Defaults to report.
input
object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: List, read, create, update or delete saved HTML reports on a board.
Use when: Use for a persistent board report rather than an ephemeral answer.
Do not use when: Do not use to compute time totals or to edit ordinary pages; use hypertask_time report or hypertask_pages instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. Identifiers and required fields depend on the selected action; writes keep existing authorization and confirmation rules.
Parameter
Type
Required
Description
action
list | get | create | update | delete
Yes
Operation to perform on this resource.
input
object | object | object | object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: Search tickets by text and filters, retrieve semantic document chunks, find related tickets, or search product help.
Use when: Use tasks for keyword discovery, semantic for relevant evidence passages, related for neighbors of a ticket, and help for product instructions.
Do not use when: Do not use when you already know a ticket number and need its current fields; use hypertask_tasks get instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. query is plain text; tasks supports the existing search operators and structured filter, semantic supports metadata_filters, related accepts a ticket identifier or text. Returned semantic evidence keeps passage text and source references.
Parameter
Type
Required
Description
action
tasks | semantic | related | help
Yes
Operation to perform on this resource.
input
object | object | object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: Mint an account MCP token or revoke signed account MCP tokens.
Use when: Use only when a human explicitly requests account credential administration.
Do not use when: Do not use for managed agent credentials; creating or revoking an identity uses hypertask_agents instead.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. expires_in_days sets token lifetime; revoke uses token or revoke_all. Mint returns a credential once; store it securely and never quote it in conversation.
Parameter
Type
Required
Description
action
mint | revoke
Yes
Operation to perform on this resource.
input
object | object
No
Action-specific fields documented in the corresponding input schema. Fields depend on action; use the variant whose title is the action.
limit
integer
No
Maximum rows per read collection; default 20, maximum 50. Defaults to 20.
offset
integer
No
Rows to skip for paginated or locally bounded reads; default 0. Prefer an input cursor when supplied. Defaults to 0.
page
integer
No
1-based page number; page 1 is the first page. An offset that equals the page start is accepted; any other offset is refused.
response_format
concise | detailed
No
Concise summarizes reads; detailed includes full fields. Writes always retain their receipts. Defaults to concise.
Does: Read or manage a managed agent’s signed outbound webhook and test or replay deliveries.
Use when: Use to configure an agent callback or diagnose a specific failed delivery.
Do not use when: Do not use for generic HTTP requests or agent creation; callbacks must belong to the authenticated agent or an owned agent.
Parameters and caveats: action selects an operation; input contains only that operation’s documented fields. response_format defaults to concise for reads; detailed retains full fields. limit defaults to 20 (maximum 50) and offset defaults to 0 for bounded read results; use returned pagination or narrower filters. agent_id=self means the authenticated managed agent; humans supply an owned agent id. Test and replay send real deliveries, and configure or rotate returns a signing secret once.
