Messages
Create a Message
Send a structured list of input messages with text and/or image content, and the model will generate the next message in the conversation.
Models
Base64ImageSource object{ type: "base64", data, media_type }
data: string
media_type: "image/jpeg" or "image/png" or "image/gif" or "image/webp"
Base64PDFSource object{ type: "base64", data, media_type }
data: string
BashCodeExecutionOutputBlock object{ type: "bash_code_execution_output", file_id }
type: "bash_code_execution_output"
BashCodeExecutionOutputBlockParam object{ type: "bash_code_execution_output", file_id }
BashCodeExecutionResultBlock object{ type: "bash_code_execution_result", content, return_code, 2 more }
type: "bash_code_execution_result"
content: array of BashCodeExecutionOutputBlock { type: "bash_code_execution_output", file_id }
type: "bash_code_execution_output"
BashCodeExecutionResultBlockParam object{ type: "bash_code_execution_result", content, return_code, 2 more }
content: array of BashCodeExecutionOutputBlockParam { type: "bash_code_execution_output", file_id }
BashCodeExecutionToolResultBlock object{ type: "bash_code_execution_tool_result", content, tool_use_id }
BashCodeExecutionToolResultBlockParam object{ type: "bash_code_execution_tool_result", content, tool_use_id, cache_control }
BashCodeExecutionToolResultError object{ type: "bash_code_execution_tool_result_error", error_code }
type: "bash_code_execution_tool_result_error"
error_code: BashCodeExecutionToolResultErrorCode
BashCodeExecutionToolResultErrorCode = "invalid_tool_input" or "unavailable" or "too_many_requests" or 2 more
BashCodeExecutionToolResultErrorParam object{ type: "bash_code_execution_tool_result_error", error_code }
error_code: BashCodeExecutionToolResultErrorCode
BrowserClickTarget = BrowserCoordinateTarget or BrowserRefTargetWhere to act: either a viewport coordinate or an element reference.
Where to act: either a viewport coordinate or an element reference.
BrowserCoordinateTarget object{ type: "coordinate", x, y }A point in the browser viewport, in viewport pixels (the same frame as a
full-viewport screenshot).
A point in the browser viewport, in viewport pixels (the same frame as a full-viewport screenshot).
x: numberPixels from the left edge of the viewport.
Pixels from the left edge of the viewport.
y: numberPixels from the top edge of the viewport.
Pixels from the top edge of the viewport.
BrowserRefTarget object{ type: "ref", ref }An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element reference (e.g. "ref_7") returned by a prior read_page or find result.
BrowserCloseTabConfig object{ defer_loading, enabled }close_tab's config overrides.
close_tab's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserCloseTabInput object{ tab_id }Close the tab with the given tab_id.
Close the tab with the given tab_id.
The tab to close.
BrowserCloseTabToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserCoordinateTarget object{ type: "coordinate", x, y }A point in the browser viewport, in viewport pixels (the same frame as a
full-viewport screenshot).
A point in the browser viewport, in viewport pixels (the same frame as a full-viewport screenshot).
x: numberPixels from the left edge of the viewport.
Pixels from the left edge of the viewport.
y: numberPixels from the top edge of the viewport.
Pixels from the top edge of the viewport.
BrowserDoubleClickConfig object{ defer_loading, enabled }double_click's config overrides.
double_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserDoubleClickInput object{ target, modifiers, tab_id }Double left-click at a viewport coordinate or on an element by reference.
Double left-click at a viewport coordinate or on an element by reference.
BrowserDoubleClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserFileUploadConfig object{ defer_loading, enabled }file_upload's config overrides.
file_upload's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserFileUploadInput object{ target, document_ids, paths, tab_id }Set the value of a file-input element to one or more files. The target must be an
element reference; at least one of paths or document_ids is required.
Set the value of a file-input element to one or more files. The target must be an element reference; at least one of paths or document_ids is required.
target: BrowserRefTarget { type: "ref", ref }An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element reference (e.g. "ref_7") returned by a prior read_page or find result.
document_ids: optional array of string or nullReferences to files the harness has staged, for deployments where the browser executor cannot read the caller's filesystem.
References to files the harness has staged, for deployments where the browser executor cannot read the caller's filesystem.
paths: optional array of string or nullFile paths on the browser executor's filesystem.
File paths on the browser executor's filesystem.
Tab to act on. Defaults to the active tab when omitted.
BrowserFileUploadToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserFindConfig object{ defer_loading, enabled }find's config overrides.
find's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserFindInput object{ query, tab_id }Find elements matching a natural-language description (e.g. "search bar", "add to
cart button") and return up to 20 matches with element references.
Find elements matching a natural-language description (e.g. "search bar", "add to cart button") and return up to 20 matches with element references.
Natural-language description of the element(s) to find.
Tab to act on. Defaults to the active tab when omitted.
BrowserFindToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserFormInputConfig object{ defer_loading, enabled }form_input's config overrides.
form_input's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserFormInputInput object{ target, value, tab_id }Set the value of a form element (input, textarea, select, checkbox). Use a
boolean for checkboxes, an option value or text for selects.
Set the value of a form element (input, textarea, select, checkbox). Use a boolean for checkboxes, an option value or text for selects.
target: BrowserRefTarget { type: "ref", ref }An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element reference (e.g. "ref_7") returned by a prior read_page or find result.
value: BrowserFormInputValueThe value to set.
The value to set.
Tab to act on. Defaults to the active tab when omitted.
BrowserFormInputToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserFormInputValue = string or number or boolean
BrowserGetPageTextConfig object{ defer_loading, enabled }get_page_text's config overrides.
get_page_text's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserGetPageTextInput object{ tab_id }Return the page's visible text content as plain text, prioritizing article
content. Suited to articles, documentation, and other text-heavy pages.
Return the page's visible text content as plain text, prioritizing article content. Suited to articles, documentation, and other text-heavy pages.
Tab to act on. Defaults to the active tab when omitted.
BrowserGetPageTextToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserHoldKeyConfig object{ defer_loading, enabled }hold_key's config overrides.
hold_key's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserHoldKeyInput object{ duration, text, tab_id }Hold a key or key chord down for a duration, then release it. Uses the same key
names and "+" chord syntax as the key action.
Hold a key or key chord down for a duration, then release it. Uses the same key names and "+" chord syntax as the key action.
duration: numberSeconds to hold the key down (maximum 30).
Seconds to hold the key down (maximum 30).
The key or chord to hold.
Tab to act on. Defaults to the active tab when omitted.
BrowserHoldKeyToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserHoverConfig object{ defer_loading, enabled }hover's config overrides.
hover's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserHoverInput object{ target, tab_id }Move the cursor to a coordinate or element without clicking.
Move the cursor to a coordinate or element without clicking.
BrowserHoverToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserJavascriptExecConfig object{ defer_loading, enabled }javascript_exec's config overrides.
javascript_exec's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserJavascriptExecInput object{ text, tab_id }Execute JavaScript in the page context and return the value of the last
expression. The code runs with access to the DOM, window, and page variables.
Write the expression you want evaluated — do NOT use return.
Execute JavaScript in the page context and return the value of the last
expression. The code runs with access to the DOM, window, and page variables.
Write the expression you want evaluated — do NOT use return.
JavaScript to execute in the page context.
Tab to act on. Defaults to the active tab when omitted.
BrowserJavascriptExecToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserKeyConfig object{ defer_loading, enabled }key's config overrides.
key's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserKeyInput object{ text, repeat, tab_id }Press a key or key chord. Use "+" to combine modifiers with a key (e.g. "ctrl+a",
"cmd+shift+p") and space to sequence presses (e.g. "Backspace Backspace Delete").
Common names like "Return", "Tab", "Escape", "BackSpace" are supported.
Press a key or key chord. Use "+" to combine modifiers with a key (e.g. "ctrl+a", "cmd+shift+p") and space to sequence presses (e.g. "Backspace Backspace Delete"). Common names like "Return", "Tab", "Escape", "BackSpace" are supported.
The key, chord, or space-separated sequence to press.
repeat: optional number or nullNumber of times to repeat. Default 1.
Number of times to repeat. Default 1.
Tab to act on. Defaults to the active tab when omitted.
BrowserKeyToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserLeftClickConfig object{ defer_loading, enabled }left_click's config overrides.
left_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserLeftClickDragConfig object{ defer_loading, enabled }left_click_drag's config overrides.
left_click_drag's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserLeftClickDragInput object{ from, target, tab_id }Press at from, drag to target, release. Both must be coordinate targets.
Press at from, drag to target, release. Both must be coordinate targets.
BrowserLeftClickDragToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserLeftClickInput object{ target, modifiers, tab_id }Left-click at a viewport coordinate or on an element by reference.
Left-click at a viewport coordinate or on an element by reference.
BrowserLeftClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserLeftMouseDownConfig object{ defer_loading, enabled }left_mouse_down's config overrides.
left_mouse_down's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserLeftMouseDownInput object{ target, tab_id }Press and hold the left mouse button at a viewport coordinate. Pair with
left_mouse_up to perform a custom drag.
Press and hold the left mouse button at a viewport coordinate. Pair with left_mouse_up to perform a custom drag.
target: BrowserCoordinateTarget { type: "coordinate", x, y }A point in the browser viewport, in viewport pixels (the same frame as a
full-viewport screenshot).
A point in the browser viewport, in viewport pixels (the same frame as a full-viewport screenshot).
x: numberPixels from the left edge of the viewport.
Pixels from the left edge of the viewport.
y: numberPixels from the top edge of the viewport.
Pixels from the top edge of the viewport.
Tab to act on. Defaults to the active tab when omitted.
BrowserLeftMouseDownToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserLeftMouseUpConfig object{ defer_loading, enabled }left_mouse_up's config overrides.
left_mouse_up's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserLeftMouseUpInput object{ target, tab_id }Release the left mouse button at a viewport coordinate.
Release the left mouse button at a viewport coordinate.
target: BrowserCoordinateTarget { type: "coordinate", x, y }A point in the browser viewport, in viewport pixels (the same frame as a
full-viewport screenshot).
A point in the browser viewport, in viewport pixels (the same frame as a full-viewport screenshot).
x: numberPixels from the left edge of the viewport.
Pixels from the left edge of the viewport.
y: numberPixels from the top edge of the viewport.
Pixels from the top edge of the viewport.
Tab to act on. Defaults to the active tab when omitted.
BrowserLeftMouseUpToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserListTabsConfig object{ defer_loading, enabled }list_tabs's config overrides.
list_tabs's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
List all open tabs with each tab's tab_id, title, and URL.
BrowserListTabsToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserMiddleClickConfig object{ defer_loading, enabled }middle_click's config overrides.
middle_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserMiddleClickInput object{ target, modifiers, tab_id }Middle-click at a viewport coordinate or on an element by reference.
Middle-click at a viewport coordinate or on an element by reference.
BrowserMiddleClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserMouseMoveConfig object{ defer_loading, enabled }mouse_move's config overrides.
mouse_move's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserMouseMoveInput object{ target, tab_id }Move the pointer to a viewport coordinate without clicking.
Move the pointer to a viewport coordinate without clicking.
target: BrowserCoordinateTarget { type: "coordinate", x, y }A point in the browser viewport, in viewport pixels (the same frame as a
full-viewport screenshot).
A point in the browser viewport, in viewport pixels (the same frame as a full-viewport screenshot).
x: numberPixels from the left edge of the viewport.
Pixels from the left edge of the viewport.
y: numberPixels from the top edge of the viewport.
Pixels from the top edge of the viewport.
Tab to act on. Defaults to the active tab when omitted.
BrowserMouseMoveToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserNavigateConfig object{ defer_loading, enabled }navigate's config overrides.
navigate's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserNavigateInput object{ url, tab_id }Navigate to a URL, or go back/forward/reload in history. The protocol may be
omitted (defaults to https://).
Navigate to a URL, or go back/forward/reload in history. The protocol may be omitted (defaults to https://).
The URL to navigate to, or "back" / "forward" / "reload" for history navigation.
Tab to act on. Defaults to the active tab when omitted.
BrowserNavigateToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserNewTabConfig object{ defer_loading, enabled }new_tab's config overrides.
new_tab's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
Open a new empty tab and return its tab_id.
BrowserNewTabToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserReadConsoleConfig object{ defer_loading, enabled }read_console's config overrides.
read_console's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserReadConsoleInput object{ tab_id }Return console output (log entries, errors, warnings) accumulated since the
driver attached to the tab and since the last read, one line per entry. An empty
result does not mean no traffic for a tab that predates attach.
Return console output (log entries, errors, warnings) accumulated since the driver attached to the tab and since the last read, one line per entry. An empty result does not mean no traffic for a tab that predates attach.
Tab to act on. Defaults to the active tab when omitted.
BrowserReadConsoleToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserReadNetworkConfig object{ defer_loading, enabled }read_network's config overrides.
read_network's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserReadNetworkInput object{ tab_id }Return the network requests (method, URL, status, MIME type, timing) recorded
since the driver attached to the tab and since the last read, one line per entry.
An empty result does not mean no traffic for a tab that predates attach.
Return the network requests (method, URL, status, MIME type, timing) recorded since the driver attached to the tab and since the last read, one line per entry. An empty result does not mean no traffic for a tab that predates attach.
Tab to act on. Defaults to the active tab when omitted.
BrowserReadNetworkToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserReadPageConfig object{ defer_loading, enabled }read_page's config overrides.
read_page's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserReadPageFilter = "all" or "interactive"
BrowserReadPageInput object{ depth, filter, ref, tab_id }Return a structured accessibility tree of the page (or the subtree rooted at
ref), with element references like [ref_7] that can be used as targets on later
actions. Output is capped at 50,000 characters — narrow with ref or a smaller
depth when exceeded.
Return a structured accessibility tree of the page (or the subtree rooted at
ref), with element references like [ref_7] that can be used as targets on later
actions. Output is capped at 50,000 characters — narrow with ref or a smaller
depth when exceeded.
depth: optional number or nullMaximum tree depth. Default 15.
Maximum tree depth. Default 15.
filter: optional BrowserReadPageFilter or nullWhich elements to include. Omitted: every visible element. "interactive": interactive elements only. "all": additionally includes off-viewport elements.
Which elements to include. Omitted: every visible element. "interactive": interactive elements only. "all": additionally includes off-viewport elements.
Element reference to read a subtree from. Omit to read from the page root.
Tab to act on. Defaults to the active tab when omitted.
BrowserReadPageToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserRefTarget object{ type: "ref", ref }An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element reference (e.g. "ref_7") returned by a prior read_page or find result.
BrowserRightClickConfig object{ defer_loading, enabled }right_click's config overrides.
right_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserRightClickInput object{ target, modifiers, tab_id }Right-click at a viewport coordinate or on an element by reference.
Right-click at a viewport coordinate or on an element by reference.
BrowserRightClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserScreenshotConfig object{ defer_loading, enabled }screenshot's config overrides.
screenshot's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserScreenshotInput object{ tab_id }Capture the current browser viewport.
Capture the current browser viewport.
Tab to act on. Defaults to the active tab when omitted.
BrowserScreenshotToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserScrollConfig object{ defer_loading, enabled }scroll's config overrides.
scroll's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserScrollDirection = "up" or "down" or "left" or "right"
BrowserScrollInput object{ scroll_direction, target, scroll_amount, tab_id }Scroll at a viewport position. target must be a coordinate target.
Scroll at a viewport position. target must be a coordinate target.
scroll_direction: BrowserScrollDirection
target: BrowserCoordinateTarget { type: "coordinate", x, y }A point in the browser viewport, in viewport pixels (the same frame as a
full-viewport screenshot).
A point in the browser viewport, in viewport pixels (the same frame as a full-viewport screenshot).
x: numberPixels from the left edge of the viewport.
Pixels from the left edge of the viewport.
y: numberPixels from the top edge of the viewport.
Pixels from the top edge of the viewport.
scroll_amount: optional number or nullScroll-wheel notches (1–10). Default 3.
Scroll-wheel notches (1–10). Default 3.
Tab to act on. Defaults to the active tab when omitted.
BrowserScrollToConfig object{ defer_loading, enabled }scroll_to's config overrides.
scroll_to's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserScrollToInput object{ target, tab_id }Scroll an element into view.
Scroll an element into view.
target: BrowserRefTarget { type: "ref", ref }An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element on the page, identified by a reference from a prior read_page or
find result. References are scoped to the tab that produced them and become
stale after navigation or a major re-render.
An element reference (e.g. "ref_7") returned by a prior read_page or find result.
Tab to act on. Defaults to the active tab when omitted.
BrowserScrollToToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserScrollToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserStateBlockParam object{ type: "browser_state", tabs, cache_control, state_changes }The caller's browser state after a browser toolset member call —
the full inventory of open tabs, which tab is active, and any side
effects (tabs opened, download state changes) the call produced.
The caller's browser state after a browser toolset member call — the full inventory of open tabs, which tab is active, and any side effects (tabs opened, download state changes) the call produced.
At most one per tool_result, only on a non-error result answering a
browser toolset member tool_use. The server renders the
model-visible text from it; the model never sees the raw fields.
BrowserStateChange = BrowserStateChangeTabOpened or BrowserStateChangeDownloadStarted or BrowserStateChangeDownloadCompleted or BrowserStateChangeDownloadFailed
BrowserStateChangeDownloadCompleted object{ type: "download_completed", download_id, url, 2 more }A file download that finished during this call, reported with the
same download_id as its download_started — or without a prior
download_started, when the download finished during the call that
started it (at most one state change per download_id per result).
A file download that finished during this call, reported with the
same download_id as its download_started — or without a prior
download_started, when the download finished during the call that
started it (at most one state change per download_id per result).
download_id: stringThe caller-assigned identifier for this download, stable across the state changes reporting it.
The caller-assigned identifier for this download, stable across the state changes reporting it.
url: stringThe final post-redirect URL the download was served from.
The final post-redirect URL the download was served from.
path: optional string or nullWhere the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
Where the executor saved the file, on the executor's filesystem. Only included when another tool in the same environment can read the file at that path.
size_bytes: optional number or nullThe completed download's size.
The completed download's size.
BrowserStateChangeDownloadFailed object{ type: "download_failed", download_id, url, error }A file download that failed — or was cancelled — during this call.
A file download that failed — or was cancelled — during this call.
download_id: stringThe caller-assigned identifier for this download, stable across the state changes reporting it.
The caller-assigned identifier for this download, stable across the state changes reporting it.
url: stringThe final post-redirect URL the download was served from.
The final post-redirect URL the download was served from.
error: optional string or nullThe failure or cancellation detail, when known.
The failure or cancellation detail, when known.
BrowserStateChangeDownloadStarted object{ type: "download_started", download_id, url }A file download that started during this call.
A file download that started during this call.
download_id: stringThe caller-assigned identifier for this download, stable across the state changes reporting it.
The caller-assigned identifier for this download, stable across the state changes reporting it.
url: stringThe final post-redirect URL the download was served from.
The final post-redirect URL the download was served from.
BrowserStateChangeTabOpened object{ type: "tab_opened", tab_id }A tab this call's execution opened that remains open at its end —
the creation delta of the tabs inventory, not an event log.
A tab this call's execution opened that remains open at its end —
the creation delta of the tabs inventory, not an event log.
Carries only the tab_id; the tab's title and url live on its
tabs entry, which must include the same tab_id. A tab opened
during a failed call gets no deferred tab_opened; it simply appears
in the next result's tabs inventory.
tab_id: stringThe tab_id of the opened tab, present in tabs.
The tab_id of the opened tab, present in tabs.
BrowserStateTabEntry object{ tab_id, title, url, active }One open browser tab reported in a browser_state block's tabs
inventory.
One open browser tab reported in a browser_state block's tabs
inventory.
tab_id is the caller-assigned identifier for the tab; title and
url describe the page the tab is currently showing and may be empty
strings (a blank tab legitimately has both empty). active marks the
tab that is active after this call; whenever tabs is non-empty,
exactly one entry is marked.
tab_id: stringThe caller-assigned identifier for this tab, unique within the inventory.
The caller-assigned identifier for this tab, unique within the inventory.
title: stringThe title of the page the tab is showing. May be empty.
The title of the page the tab is showing. May be empty.
url: stringThe URL of the page the tab is showing. May be empty.
The URL of the page the tab is showing. May be empty.
Whether this tab is the active tab after this call. Whenever tabs is non-empty, exactly one entry is marked active: true.
BrowserSwitchTabConfig object{ defer_loading, enabled }switch_tab's config overrides.
switch_tab's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserSwitchTabInput object{ tab_id }Make the tab with the given tab_id the active tab — the tab that actions without
a tab_id apply to.
Make the tab with the given tab_id the active tab — the tab that actions without a tab_id apply to.
The tab to switch to.
BrowserSwitchTabToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserToolUseBlock = BrowserNavigateToolUseBlock or BrowserListTabsToolUseBlock or BrowserNewTabToolUseBlock or 28 more
BrowserToolset20260801 object{ type: "browser_toolset_20260801", cache_control, configs }The browser toolset: a single tools[] entry (carrying no
name) that declares the browser tool family. The model is served
the family's tool with any members disabled via configs removed
from its schema.
The browser toolset: a single tools[] entry (carrying no
name) that declares the browser tool family. The model is served
the family's tool with any members disabled via configs removed
from its schema.
BrowserToolsetConfigs object{ type, close_tab, double_click, 28 more }Per-member configuration for browser_toolset_20260801: one
optional field per member tool, keyed by the member name — the same
name the member's tool_use blocks carry. Every member is an
accepted key, and a member's defaults apply wherever its key is
absent. Unknown keys are rejected: the field set is this toolset
version's complete member set.
Per-member configuration for browser_toolset_20260801: one
optional field per member tool, keyed by the member name — the same
name the member's tool_use blocks carry. Every member is an
accepted key, and a member's defaults apply wherever its key is
absent. Unknown keys are rejected: the field set is this toolset
version's complete member set.
BrowserTripleClickConfig object{ defer_loading, enabled }triple_click's config overrides.
triple_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserTripleClickInput object{ target, modifiers, tab_id }Triple left-click at a viewport coordinate or on an element by reference
(typically selects a line or paragraph).
Triple left-click at a viewport coordinate or on an element by reference (typically selects a line or paragraph).
BrowserTripleClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserTypeConfig object{ defer_loading, enabled }type's config overrides.
type's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserTypeInput object{ text, tab_id }Type a literal string at the current focus.
Type a literal string at the current focus.
The text to type.
Tab to act on. Defaults to the active tab when omitted.
BrowserTypeToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserWaitConfig object{ defer_loading, enabled }wait's config overrides.
wait's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserWaitInput object{ duration, tab_id }Pause for the given duration.
Pause for the given duration.
duration: numberSeconds to wait (maximum 30).
Seconds to wait (maximum 30).
Tab to act on. Defaults to the active tab when omitted.
BrowserWaitToolUseBlock object{ type: "tool_use", id, caller, 3 more }
BrowserZoomConfig object{ defer_loading, enabled }zoom's config overrides.
zoom's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
BrowserZoomInput object{ region, tab_id }Return a cropped screenshot of the given viewport region, scaled up for closer
inspection — useful for small icons, buttons, or text. Coordinates are in the
same viewport-pixel space as a full screenshot.
Return a cropped screenshot of the given viewport region, scaled up for closer inspection — useful for small icons, buttons, or text. Coordinates are in the same viewport-pixel space as a full screenshot.
region: array of number[x0, y0, x1, y1] in viewport pixels.
[x0, y0, x1, y1] in viewport pixels.
Tab to act on. Defaults to the active tab when omitted.
BrowserZoomToolUseBlock object{ type: "tool_use", id, caller, 3 more }
CacheControlEphemeral object{ type: "ephemeral", ttl }
ttl: optional "5m" or "1h"The time-to-live for the cache control breakpoint.
The time-to-live for the cache control breakpoint.
This may be one the following values:
5m: 5 minutes1h: 1 hour
Defaults to 5m. See prompt caching pricing for details.
CacheCreation object{ ephemeral_1h_input_tokens, ephemeral_5m_input_tokens }
ephemeral_1h_input_tokens: numberThe number of input tokens used to create the 1 hour cache entry.
The number of input tokens used to create the 1 hour cache entry.
ephemeral_5m_input_tokens: numberThe number of input tokens used to create the 5 minute cache entry.
The number of input tokens used to create the 5 minute cache entry.
CacheMissMessagesChanged object{ type: "messages_changed", cache_missed_input_tokens }
type: "messages_changed"
Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.
CacheMissModelChanged object{ type: "model_changed", cache_missed_input_tokens }
type: "model_changed"
Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.
CacheMissPreviousMessageNotFound object{ type: "previous_message_not_found" }
type: "previous_message_not_found"
CacheMissReason = CacheMissModelChanged or CacheMissSystemChanged or CacheMissToolsChanged or 3 more
CacheMissSystemChanged object{ type: "system_changed", cache_missed_input_tokens }
type: "system_changed"
Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.
CacheMissToolsChanged object{ type: "tools_changed", cache_missed_input_tokens }
type: "tools_changed"
Approximate number of input tokens that would have been read from cache had the prefix matched the previous request.
CacheMissUnavailable object{ type: "unavailable" }
type: "unavailable"
CitationCharLocation object{ type: "char_location", cited_text, document_index, 4 more }
type: "char_location"
document_index: number
start_char_index: number
CitationCharLocationParam object{ type: "char_location", cited_text, document_index, 3 more }
document_index: number
document_title: string or null
start_char_index: number
CitationContentBlockLocation object{ type: "content_block_location", cited_text, document_index, 4 more }
type: "content_block_location"
cited_text: stringThe full text of the cited block range, concatenated.
The full text of the cited block range, concatenated.
Always equals the contents of content[start_block_index:end_block_index] joined together. The text block is the minimal citable unit; this field is never a substring of a single block. Not counted toward output tokens, and not counted toward input tokens when sent back in subsequent turns.
document_index: number
end_block_index: numberExclusive 0-based end index of the cited block range in the source's content array.
Exclusive 0-based end index of the cited block range in the source's content array.
Always greater than start_block_index; a single-block citation has end_block_index = start_block_index + 1.
start_block_index: number0-based index of the first cited block in the source's content array.
0-based index of the first cited block in the source's content array.
CitationContentBlockLocationParam object{ type: "content_block_location", cited_text, document_index, 3 more }
cited_text: stringThe full text of the cited block range, concatenated.
The full text of the cited block range, concatenated.
Always equals the contents of content[start_block_index:end_block_index] joined together. The text block is the minimal citable unit; this field is never a substring of a single block. Not counted toward output tokens, and not counted toward input tokens when sent back in subsequent turns.
document_index: number
document_title: string or null
end_block_index: numberExclusive 0-based end index of the cited block range in the source's content array.
Exclusive 0-based end index of the cited block range in the source's content array.
Always greater than start_block_index; a single-block citation has end_block_index = start_block_index + 1.
start_block_index: number0-based index of the first cited block in the source's content array.
0-based index of the first cited block in the source's content array.
CitationPageLocation object{ type: "page_location", cited_text, document_index, 4 more }
type: "page_location"
document_index: number
start_page_number: number
CitationPageLocationParam object{ type: "page_location", cited_text, document_index, 3 more }
document_index: number
document_title: string or null
start_page_number: number
CitationSearchResultLocationParam object{ type: "search_result_location", cited_text, end_block_index, 4 more }
cited_text: stringThe full text of the cited block range, concatenated.
The full text of the cited block range, concatenated.
Always equals the contents of content[start_block_index:end_block_index] joined together. The text block is the minimal citable unit; this field is never a substring of a single block. Not counted toward output tokens, and not counted toward input tokens when sent back in subsequent turns.
end_block_index: numberExclusive 0-based end index of the cited block range in the source's content array.
Exclusive 0-based end index of the cited block range in the source's content array.
Always greater than start_block_index; a single-block citation has end_block_index = start_block_index + 1.
search_result_index: number0-based index of the cited search result among all search_result content blocks in the request, in the order they appear across messages and tool results.
0-based index of the cited search result among all search_result content blocks in the request, in the order they appear across messages and tool results.
Counted separately from document_index; server-side web search results are not included in this count.
start_block_index: number0-based index of the first cited block in the source's content array.
0-based index of the first cited block in the source's content array.
CitationWebSearchResultLocationParam object{ type: "web_search_result_location", cited_text, encrypted_index, 2 more }
title: string or null
url: string
CitationsConfig object{ enabled }
enabled: boolean
CitationsConfigParam object{ enabled }
CitationsDelta object{ type: "citations_delta", citation }
CitationsSearchResultLocation object{ type: "search_result_location", cited_text, end_block_index, 4 more }
type: "search_result_location"
cited_text: stringThe full text of the cited block range, concatenated.
The full text of the cited block range, concatenated.
Always equals the contents of content[start_block_index:end_block_index] joined together. The text block is the minimal citable unit; this field is never a substring of a single block. Not counted toward output tokens, and not counted toward input tokens when sent back in subsequent turns.
end_block_index: numberExclusive 0-based end index of the cited block range in the source's content array.
Exclusive 0-based end index of the cited block range in the source's content array.
Always greater than start_block_index; a single-block citation has end_block_index = start_block_index + 1.
search_result_index: number0-based index of the cited search result among all search_result content blocks in the request, in the order they appear across messages and tool results.
0-based index of the cited search result among all search_result content blocks in the request, in the order they appear across messages and tool results.
Counted separately from document_index; server-side web search results are not included in this count.
start_block_index: number0-based index of the first cited block in the source's content array.
0-based index of the first cited block in the source's content array.
CitationsWebSearchResultLocation object{ type: "web_search_result_location", cited_text, encrypted_index, 2 more }
type: "web_search_result_location"
title: string or null
CodeExecutionOutputBlock object{ type: "code_execution_output", file_id }
type: "code_execution_output"
CodeExecutionOutputBlockParam object{ type: "code_execution_output", file_id }
CodeExecutionResultBlock object{ type: "code_execution_result", content, return_code, 2 more }
type: "code_execution_result"
content: array of CodeExecutionOutputBlock { type: "code_execution_output", file_id }
type: "code_execution_output"
CodeExecutionResultBlockParam object{ type: "code_execution_result", content, return_code, 2 more }
content: array of CodeExecutionOutputBlockParam { type: "code_execution_output", file_id }
CodeExecutionTool20250522 object{ type: "code_execution_20250522", name, allowed_callers, 3 more }
CodeExecutionTool20250825 object{ type: "code_execution_20250825", name, allowed_callers, 3 more }
CodeExecutionTool20260120 object{ type: "code_execution_20260120", name, allowed_callers, 3 more }Code execution tool with REPL state persistence (daemon mode + gVisor checkpoint).
Code execution tool with REPL state persistence (daemon mode + gVisor checkpoint).
CodeExecutionTool20260521 object{ type: "code_execution_20260521", name, allowed_callers, 3 more }Code execution tool with REPL state persistence.
Code execution tool with REPL state persistence.
CodeExecutionToolResultBlock object{ type: "code_execution_tool_result", content, tool_use_id }
type: "code_execution_tool_result"
content: CodeExecutionToolResultBlockContent
tool_use_id: string
CodeExecutionToolResultBlockContent = CodeExecutionToolResultError or CodeExecutionResultBlock or EncryptedCodeExecutionResultBlock
CodeExecutionToolResultBlockParam object{ type: "code_execution_tool_result", content, tool_use_id, cache_control }
tool_use_id: string
cache_control: optional CacheControlEphemeral { type: "ephemeral", ttl } or nullCreate a cache control breakpoint at this content block.
Create a cache control breakpoint at this content block.
ttl: optional "5m" or "1h"The time-to-live for the cache control breakpoint.
The time-to-live for the cache control breakpoint.
This may be one the following values:
5m: 5 minutes1h: 1 hour
Defaults to 5m. See prompt caching pricing for details.
CodeExecutionToolResultBlockParamContent = CodeExecutionToolResultErrorParam or CodeExecutionResultBlockParam or EncryptedCodeExecutionResultBlockParam
CodeExecutionToolResultError object{ type: "code_execution_tool_result_error", error_code }
type: "code_execution_tool_result_error"
error_code: CodeExecutionToolResultErrorCode
CodeExecutionToolResultErrorCode = "invalid_tool_input" or "unavailable" or "too_many_requests" or "execution_time_exceeded"
CodeExecutionToolResultErrorParam object{ type: "code_execution_tool_result_error", error_code }
error_code: CodeExecutionToolResultErrorCode
ComputerCursorPositionConfig object{ defer_loading, enabled }cursor_position's config overrides.
cursor_position's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
Get the current (x, y) pixel coordinate of the cursor.
ComputerCursorPositionToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerDoubleClickConfig object{ defer_loading, enabled }double_click's config overrides.
double_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerDoubleClickInput object{ coordinate, text }Double-click the left mouse button at the specified (x, y) pixel coordinate, or
the current cursor position if coordinate is omitted.
Double-click the left mouse button at the specified (x, y) pixel coordinate, or
the current cursor position if coordinate is omitted.
coordinate: optional array of number or null(x, y): x pixels from the left edge, y pixels from the top edge.
(x, y): x pixels from the left edge, y pixels from the top edge.
Optional key combination to hold down during this action (e.g. "ctrl", "shift", "ctrl+shift").
ComputerDoubleClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerHoldKeyConfig object{ defer_loading, enabled }hold_key's config overrides.
hold_key's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerHoldKeyInput object{ duration, text }Hold down a key or key-combination for a specified duration. Uses the same key
syntax as key.
Hold down a key or key-combination for a specified duration. Uses the same key
syntax as key.
duration: numberDuration to hold the key, in seconds.
Duration to hold the key, in seconds.
The key or key-combination to hold.
ComputerHoldKeyToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerKeyConfig object{ defer_loading, enabled }key's config overrides.
key's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerKeyInput object{ text, repeat }Press a key or key-combination on the keyboard. Use "+" to combine modifiers with
a key (e.g. "ctrl+s", "alt+Tab", "ctrl+shift+Escape"). Key names are
case-insensitive; common names like "Return", "Tab", "Escape", "Up", "Down",
"Left", "Right", "Home", "End", "Page_Up", "Page_Down", "Delete", "BackSpace" are
supported.
Press a key or key-combination on the keyboard. Use "+" to combine modifiers with a key (e.g. "ctrl+s", "alt+Tab", "ctrl+shift+Escape"). Key names are case-insensitive; common names like "Return", "Tab", "Escape", "Up", "Down", "Left", "Right", "Home", "End", "Page_Up", "Page_Down", "Delete", "BackSpace" are supported.
The key or key-combination to press.
repeat: optional number or nullNumber of times to repeat the key press. Default is 1.
Number of times to repeat the key press. Default is 1.
ComputerKeyToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerLeftClickConfig object{ defer_loading, enabled }left_click's config overrides.
left_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerLeftClickDragConfig object{ defer_loading, enabled }left_click_drag's config overrides.
left_click_drag's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerLeftClickDragInput object{ coordinate, start_coordinate, text }Click and drag the cursor from start_coordinate to coordinate.
Click and drag the cursor from start_coordinate to coordinate.
coordinate: array of number(x, y): x pixels from the left edge, y pixels from the top edge.
(x, y): x pixels from the left edge, y pixels from the top edge.
start_coordinate: array of number(x, y): x pixels from the left edge, y pixels from the top edge.
(x, y): x pixels from the left edge, y pixels from the top edge.
Optional key combination to hold down during this action (e.g. "ctrl", "shift", "ctrl+shift").
ComputerLeftClickDragToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerLeftClickInput object{ coordinate, text }Click the left mouse button at the specified (x, y) pixel coordinate, or the
current cursor position if coordinate is omitted.
Click the left mouse button at the specified (x, y) pixel coordinate, or the
current cursor position if coordinate is omitted.
coordinate: optional array of number or null(x, y): x pixels from the left edge, y pixels from the top edge.
(x, y): x pixels from the left edge, y pixels from the top edge.
Optional key combination to hold down during this action (e.g. "ctrl", "shift", "ctrl+shift").
ComputerLeftClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerLeftMouseDownConfig object{ defer_loading, enabled }left_mouse_down's config overrides.
left_mouse_down's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
Press and hold the left mouse button at the current cursor position.
ComputerLeftMouseDownToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerLeftMouseUpConfig object{ defer_loading, enabled }left_mouse_up's config overrides.
left_mouse_up's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
Release the left mouse button.
ComputerLeftMouseUpToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerMiddleClickConfig object{ defer_loading, enabled }middle_click's config overrides.
middle_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerMiddleClickInput object{ coordinate, text }Click the middle mouse button at the specified (x, y) pixel coordinate, or the
current cursor position if coordinate is omitted.
Click the middle mouse button at the specified (x, y) pixel coordinate, or the
current cursor position if coordinate is omitted.
coordinate: optional array of number or null(x, y): x pixels from the left edge, y pixels from the top edge.
(x, y): x pixels from the left edge, y pixels from the top edge.
Optional key combination to hold down during this action (e.g. "ctrl", "shift", "ctrl+shift").
ComputerMiddleClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerMouseMoveConfig object{ defer_loading, enabled }mouse_move's config overrides.
mouse_move's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerMouseMoveInput object{ coordinate }Move the cursor to a specified (x, y) pixel coordinate. Use this ONLY to hover
without clicking; otherwise use a click action directly.
Move the cursor to a specified (x, y) pixel coordinate. Use this ONLY to hover without clicking; otherwise use a click action directly.
coordinate: array of number(x, y): x pixels from the left edge, y pixels from the top edge.
(x, y): x pixels from the left edge, y pixels from the top edge.
ComputerMouseMoveToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerRightClickConfig object{ defer_loading, enabled }right_click's config overrides.
right_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerRightClickInput object{ coordinate, text }Click the right mouse button at the specified (x, y) pixel coordinate, or the
current cursor position if coordinate is omitted.
Click the right mouse button at the specified (x, y) pixel coordinate, or the
current cursor position if coordinate is omitted.
coordinate: optional array of number or null(x, y): x pixels from the left edge, y pixels from the top edge.
(x, y): x pixels from the left edge, y pixels from the top edge.
Optional key combination to hold down during this action (e.g. "ctrl", "shift", "ctrl+shift").
ComputerRightClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerScreenshotConfig object{ defer_loading, enabled }screenshot's config overrides.
screenshot's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
Take a screenshot of the screen.
ComputerScreenshotToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerScrollConfig object{ defer_loading, enabled }scroll's config overrides.
scroll's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerScrollDirection = "up" or "down" or "left" or "right"
ComputerScrollInput object{ scroll_amount, scroll_direction, coordinate, text }Scroll the screen at the specified (x, y) pixel coordinate, or the current cursor
position if coordinate is omitted. Do NOT use PageUp/PageDown to scroll.
Scroll the screen at the specified (x, y) pixel coordinate, or the current cursor
position if coordinate is omitted. Do NOT use PageUp/PageDown to scroll.
Number of 'clicks' of the scroll wheel.
scroll_direction: ComputerScrollDirection
coordinate: optional array of number or null(x, y): x pixels from the left edge, y pixels from the top edge.
(x, y): x pixels from the left edge, y pixels from the top edge.
Optional key combination to hold down during this action (e.g. "ctrl", "shift", "ctrl+shift").
ComputerScrollToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerToolUseBlock = ComputerKeyToolUseBlock or ComputerHoldKeyToolUseBlock or ComputerTypeToolUseBlock or 14 more
ComputerToolset20260801 object{ type: "computer_toolset_20260801", cache_control, configs }The computer toolset: a single tools[] entry (carrying no
name) that declares the computer tool family. The model is
served the family's tool with any members disabled via configs
removed from its schema. Every member is enabled by default, zoom
included. The single-tool options display_number and
enable_zoom are not fields of a toolset entry — it carries only
type, configs, and cache_control; zoom is controlled
via configs.zoom.enabled.
The computer toolset: a single tools[] entry (carrying no
name) that declares the computer tool family. The model is
served the family's tool with any members disabled via configs
removed from its schema. Every member is enabled by default, zoom
included. The single-tool options display_number and
enable_zoom are not fields of a toolset entry — it carries only
type, configs, and cache_control; zoom is controlled
via configs.zoom.enabled.
ComputerToolsetConfigs object{ type, cursor_position, double_click, 14 more }Per-member configuration for computer_toolset_20260801: one
optional field per member tool, keyed by the member name — the same
name the member's tool_use blocks carry. Every member is an
accepted key, and a member's defaults apply wherever its key is
absent. Unknown keys are rejected: the field set is this toolset
version's complete member set.
Per-member configuration for computer_toolset_20260801: one
optional field per member tool, keyed by the member name — the same
name the member's tool_use blocks carry. Every member is an
accepted key, and a member's defaults apply wherever its key is
absent. Unknown keys are rejected: the field set is this toolset
version's complete member set.
ComputerTripleClickConfig object{ defer_loading, enabled }triple_click's config overrides.
triple_click's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerTripleClickInput object{ coordinate, text }Triple-click the left mouse button at the specified (x, y) pixel coordinate, or
the current cursor position if coordinate is omitted.
Triple-click the left mouse button at the specified (x, y) pixel coordinate, or
the current cursor position if coordinate is omitted.
coordinate: optional array of number or null(x, y): x pixels from the left edge, y pixels from the top edge.
(x, y): x pixels from the left edge, y pixels from the top edge.
Optional key combination to hold down during this action (e.g. "ctrl", "shift", "ctrl+shift").
ComputerTripleClickToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerTypeConfig object{ defer_loading, enabled }type's config overrides.
type's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerTypeInput object{ text }Type a string of text on the keyboard.
Type a string of text on the keyboard.
The text to type.
ComputerTypeToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerWaitConfig object{ defer_loading, enabled }wait's config overrides.
wait's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerWaitInput object{ duration }Wait for a specified duration.
Wait for a specified duration.
duration: numberDuration to wait, in seconds.
Duration to wait, in seconds.
ComputerWaitToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ComputerZoomConfig object{ defer_loading, enabled }zoom's config overrides.
zoom's config overrides.
Defer loading for this member. Must resolve to the same value on every enabled member of the toolset.
Whether this member is offered to the model. Default is per member, per the toolset's documentation. A member whose enabled resolves false is withheld from the served schema.
ComputerZoomInput object{ region }Take a screenshot of a rectangular region. Region coordinates are in the
full-screenshot space (not physical display pixels). The crop is scaled up to
fill the image budget so fine details become legible.
Take a screenshot of a rectangular region. Region coordinates are in the full-screenshot space (not physical display pixels). The crop is scaled up to fill the image budget so fine details become legible.
region: array of number(x0, y0, x1, y1): The region to capture.
(x0, y0, x1, y1): The region to capture.
ComputerZoomToolUseBlock object{ type: "tool_use", id, caller, 3 more }
Container object{ id, expires_at, skills }Information about the container used in the request (for the code execution tool)
Information about the container used in the request (for the code execution tool)
Identifier for the container used in this request
expires_at: stringThe time at which the container will expire.
The time at which the container will expire.
skills: array of ContainerSkill { type, skill_id, version } or nullSkills loaded in the container
Skills loaded in the container
type: "anthropic" or "custom"Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
skill_id: stringSkill ID
Skill ID
version: stringThe resolved version: a skill version ID for custom skills.
The resolved version: a skill version ID for custom skills.
ContainerParams object{ id, skills }Container parameters with skills to be loaded.
Container parameters with skills to be loaded.
Container id
skills: optional array of SkillParams { type, skill_id, version } or nullList of skills to load in the container
List of skills to load in the container
type: "anthropic" or "custom"Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
skill_id: stringSkill ID
Skill ID
version: optional stringSkill version or 'latest' for most recent version
Skill version or 'latest' for most recent version
ContainerSkill object{ type, skill_id, version }A skill that was loaded in a container (response model).
A skill that was loaded in a container (response model).
type: "anthropic" or "custom"Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
skill_id: stringSkill ID
Skill ID
version: stringThe resolved version: a skill version ID for custom skills.
The resolved version: a skill version ID for custom skills.
ContainerUploadBlock object{ type: "container_upload", file_id }Response model for a file uploaded to the container.
Response model for a file uploaded to the container.
type: "container_upload"
ContainerUploadBlockParam object{ type: "container_upload", file_id, cache_control }A content block that represents a file to be uploaded to the container
Files uploaded via this block will be available in the container's input directory.
A content block that represents a file to be uploaded to the container Files uploaded via this block will be available in the container's input directory.
cache_control: optional CacheControlEphemeral { type: "ephemeral", ttl } or nullCreate a cache control breakpoint at this content block.
Create a cache control breakpoint at this content block.
ttl: optional "5m" or "1h"The time-to-live for the cache control breakpoint.
The time-to-live for the cache control breakpoint.
This may be one the following values:
5m: 5 minutes1h: 1 hour
Defaults to 5m. See prompt caching pricing for details.
ContentBlock = TextBlock or ThinkingBlock or RedactedThinkingBlock or 9 more
ContentBlockParam = TextBlockParam or ImageBlockParam or DocumentBlockParam or 13 more
ContentBlockSource object{ type: "content", content }
content: string or array of ContentBlockSourceContent
ContentBlockSourceContent = array of ContentBlockSourceContent
TextBlockParam object{ type: "text", text, cache_control, citations }
ImageBlockParam object{ type: "image", source, cache_control, transformations }
ContentBlockSourceContent = TextBlockParam or ImageBlockParam
TextBlockParam object{ type: "text", text, cache_control, citations }
ImageBlockParam object{ type: "image", source, cache_control, transformations }
Diagnostics object{ cache_miss_reason }Request-level diagnostics: why the prompt cache could not fully reuse
the prefix of the request named by diagnostics.previous_message_id.
Request-level diagnostics: why the prompt cache could not fully reuse
the prefix of the request named by diagnostics.previous_message_id.
cache_miss_reason: CacheMissReason or nullExplains why the prompt cache could not fully reuse the prefix from the request identified by diagnostics.previous_message_id. null means diagnosis is still pending — the response was serialized before the background comparison completed.
Explains why the prompt cache could not fully reuse the prefix from the request identified by diagnostics.previous_message_id. null means diagnosis is still pending — the response was serialized before the background comparison completed.
DiagnosticsParam object{ previous_message_id }Request-level diagnostics. Currently carries the previous response
id for prompt-cache divergence reporting.
Request-level diagnostics. Currently carries the previous response id for prompt-cache divergence reporting.
previous_message_id: optional string or nullThe id (msg_...) from this client's previous /v1/messages response. The server compares that request's prompt fingerprint against this one and returns diagnostics.cache_miss_reason when the prompt-cache prefix could not be reused. Pass null on the first turn to opt in without a prior message to compare.
The id (msg_...) from this client's previous /v1/messages response. The server compares that request's prompt fingerprint against this one and returns diagnostics.cache_miss_reason when the prompt-cache prefix could not be reused. Pass null on the first turn to opt in without a prior message to compare.
DirectCaller object{ type: "direct" }Tool invocation directly from the model.
Tool invocation directly from the model.
DocumentBlock object{ type: "document", citations, source, title }
DocumentBlockParam object{ type: "document", source, cache_control, 3 more }
EncryptedCodeExecutionResultBlock object{ type: "encrypted_code_execution_result", content, encrypted_stdout, 2 more }Code execution result with encrypted stdout for PFC + web_search results.
Code execution result with encrypted stdout for PFC + web_search results.
type: "encrypted_code_execution_result"
content: array of CodeExecutionOutputBlock { type: "code_execution_output", file_id }
type: "code_execution_output"
EncryptedCodeExecutionResultBlockParam object{ type: "encrypted_code_execution_result", content, encrypted_stdout, 2 more }Code execution result with encrypted stdout for PFC + web_search results.
Code execution result with encrypted stdout for PFC + web_search results.
content: array of CodeExecutionOutputBlockParam { type: "code_execution_output", file_id }
FileDocumentSource object{ type: "file", file_id }
FileImageSource object{ type: "file", file_id }
ImageBlockParam object{ type: "image", source, cache_control, transformations }
ImageTransformationsParam object{ oversized_image }Configures the transformations the server applies to this image before the model observes it. Each key names a condition the server transforms images for; its value selects the transformation applied. Omitted keys keep their default behavior, and an empty object is equivalent to omitting the field.
Configures the transformations the server applies to this image before the model observes it. Each key names a condition the server transforms images for; its value selects the transformation applied. Omitted keys keep their default behavior, and an empty object is equivalent to omitting the field.
oversized_image: optional "downsize" or "error"What the server does when this image exceeds the model's maximum image size. "downsize" (the default) scales the image down to fit, which changes the dimensions the model observes without telling you. "error" instead rejects the request with a 400 error naming the image's dimensions and the largest dimensions that fit, so you can scale the image deliberately — your image is never silently scaled down.
What the server does when this image exceeds the model's maximum image size. "downsize" (the default) scales the image down to fit, which changes the dimensions the model observes without telling you. "error" instead rejects the request with a 400 error naming the image's dimensions and the largest dimensions that fit, so you can scale the image deliberately — your image is never silently scaled down.
InputJSONDelta object{ type: "input_json_delta", partial_json }
type: "input_json_delta"
JSONOutputFormat object{ type: "json_schema", schema }
The JSON schema of the format
MemoryTool20250818 object{ type: "memory_20250818", name, allowed_callers, 4 more }
Message object{ type: "message", id, container, 8 more }
MessageCountTokensTool = Tool or ToolBash20250124 or CodeExecutionTool20250522 or 18 more
MessageCreateParamsContainer = ContainerParams or stringContainer identifier for reuse across requests.
Container identifier for reuse across requests.
ContainerParams object{ id, skills }Container parameters with skills to be loaded.
Container parameters with skills to be loaded.
Container id
skills: optional array of SkillParams { type, skill_id, version } or nullList of skills to load in the container
List of skills to load in the container
type: "anthropic" or "custom"Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
skill_id: stringSkill ID
Skill ID
version: optional stringSkill version or 'latest' for most recent version
Skill version or 'latest' for most recent version
MessageDeltaUsage object{ cache_creation_input_tokens, cache_read_input_tokens, input_tokens, 3 more }
MessageParam object{ content, role }
MessageTokensCount object{ input_tokens }
The total number of tokens across the provided list of messages, system prompt, and tools.
Metadata object{ user_id }
user_id: optional string or nullAn external identifier for the user who is associated with the request.
An external identifier for the user who is associated with the request.
This should be a uuid, hash value, or other opaque identifier. Anthropic may use this id to help detect abuse. Do not include any identifying information such as name, email address, or phone number.
Model = "claude-haiku-5-5" or "claude-sonnet-5-5" or "claude-fable-5-1" or 17 more or stringThe model that will complete your prompt.
The model that will complete your prompt.
See models for additional details and options.
OutputConfig object{ effort, format }
effort: optional "low" or "medium" or "high" or 2 more or nullHow much effort the model should put into its response. Higher effort levels may result in more thorough analysis but take longer.
How much effort the model should put into its response. Higher effort levels may result in more thorough analysis but take longer.
Valid values are low, medium, high, xhigh, or max.
format: optional JSONOutputFormat { type: "json_schema", schema } or nullA schema to specify Claude's output format in responses. See structured outputs
A schema to specify Claude's output format in responses. See structured outputs
The JSON schema of the format
OutputTokensDetails object{ thinking_tokens }
thinking_tokens: numberNumber of output tokens the model generated as internal reasoning, including
the thinking-block delimiter tokens.
Number of output tokens the model generated as internal reasoning, including the thinking-block delimiter tokens.
Reflects the raw reasoning the model produced, not the (possibly shorter)
summarized thinking text returned in the response body. Computed by
re-tokenizing the raw reasoning text, so it may differ from the model's exact
generation count by a small number of tokens. Always ≤ output_tokens;
output_tokens - thinking_tokens approximates the non-reasoning output.
PlainTextSource object{ type: "text", data, media_type }
RawContentBlockDelta = TextDelta or InputJSONDelta or CitationsDelta or 2 more
RawContentBlockDeltaEvent object{ type: "content_block_delta", delta, index }
type: "content_block_delta"
delta: RawContentBlockDelta
RawContentBlockStartEvent object{ type: "content_block_start", content_block, index }
RawContentBlockStopEvent object{ type: "content_block_stop", index }
type: "content_block_stop"
RawMessageDeltaEvent object{ type: "message_delta", delta, usage }
RawMessageStartEvent object{ type: "message_start", message }
type: "message_start"
message: Message { type: "message", id, container, 8 more }
RawMessageStopEvent object{ type: "message_stop" }
type: "message_stop"
RawMessageStreamEvent = RawMessageStartEvent or RawMessageDeltaEvent or RawMessageStopEvent or 3 more
RedactedThinkingBlock object{ type: "redacted_thinking", data }
type: "redacted_thinking"
data: stringThe contents of this redacted thinking block, returned when portions of the model's thinking were safety-redacted. This field is opaque and encrypted, with no readable content.
The contents of this redacted thinking block, returned when portions of the model's thinking were safety-redacted. This field is opaque and encrypted, with no readable content.
Pass redacted_thinking blocks back to the API unchanged when continuing a multi-turn conversation.
See extended thinking for details.
RedactedThinkingBlockParam object{ type: "redacted_thinking", data }
The data value of this redacted thinking block, exactly as returned by the API in a previous response. Opaque and encrypted; pass it back unchanged.
RefusalStopDetails object{ type: "refusal", category, explanation }Structured information about a refusal.
Structured information about a refusal.
type: "refusal"
category: "cyber" or "bio" or "frontier_llm" or 2 more or nullThe policy category that triggered the refusal.
The policy category that triggered the refusal.
null when the refusal doesn't map to a named category.
The request could enable cyber harm, such as malware or exploit development. Benign cybersecurity work can also trigger this category.
The request could enable biological harm, such as dangerous lab methods. Beneficial life sciences work can also trigger this category.
The request could assist the development of competing AI models, which is restricted under Anthropic's commercial terms. Benign machine learning work can also trigger this category.
The request asks the model to reproduce its internal reasoning in the response text. To get reasoning in a structured form instead, use adaptive thinking.
The request could be related to an area that was determined as harmful. Benign work might sometimes trigger this category.
explanation: string or nullHuman-readable explanation of the refusal.
Human-readable explanation of the refusal.
This text is not guaranteed to be stable. null when no explanation is available for the category.
SearchResultBlockParam object{ type: "search_result", content, source, 3 more }
ServerToolCaller object{ type: "code_execution_20250825", tool_id }Tool invocation generated by a server-side tool.
Tool invocation generated by a server-side tool.
tool_id: string
ServerToolCaller20260120 object{ type: "code_execution_20260120", tool_id }
tool_id: string
ServerToolUsage object{ web_fetch_requests, web_search_requests }
web_fetch_requests: numberThe number of web fetch tool requests.
The number of web fetch tool requests.
web_search_requests: numberThe number of web search tool requests.
The number of web search tool requests.
ServerToolUseBlock object{ type: "server_tool_use", id, caller, 2 more }
ServerToolUseBlockParam object{ type: "server_tool_use", id, input, 3 more }
SignatureDelta object{ type: "signature_delta", signature }
type: "signature_delta"
The signature for this thinking block: an opaque value used to verify that the block was generated by Claude when it is passed back to the API. Delivered in a signature_delta event just before the block's content_block_stop event.
SkillParams object{ type, skill_id, version }Specification for a skill to be loaded in a container (request model).
Specification for a skill to be loaded in a container (request model).
type: "anthropic" or "custom"Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
Type of skill - either 'anthropic' (built-in) or 'custom' (user-defined)
skill_id: stringSkill ID
Skill ID
version: optional stringSkill version or 'latest' for most recent version
Skill version or 'latest' for most recent version
StopReason = "end_turn" or "max_tokens" or "stop_sequence" or 4 more
TextBlock object{ type: "text", citations, text }
TextBlockParam object{ type: "text", text, cache_control, citations }
TextCitation = CitationCharLocation or CitationPageLocation or CitationContentBlockLocation or 2 more
TextCitationParam = CitationCharLocationParam or CitationPageLocationParam or CitationContentBlockLocationParam or 2 more
TextDelta object{ type: "text_delta", text }
type: "text_delta"
TextEditorCodeExecutionCreateResultBlock object{ type: "text_editor_code_execution_create_result", is_file_update }
type: "text_editor_code_execution_create_result"
TextEditorCodeExecutionCreateResultBlockParam object{ type: "text_editor_code_execution_create_result", is_file_update }
TextEditorCodeExecutionStrReplaceResultBlock object{ type: "text_editor_code_execution_str_replace_result", lines, new_lines, 3 more }
type: "text_editor_code_execution_str_replace_result"
TextEditorCodeExecutionStrReplaceResultBlockParam object{ type: "text_editor_code_execution_str_replace_result", lines, new_lines, 3 more }
TextEditorCodeExecutionToolResultBlock object{ type: "text_editor_code_execution_tool_result", content, tool_use_id }
TextEditorCodeExecutionToolResultBlockParam object{ type: "text_editor_code_execution_tool_result", content, tool_use_id, cache_control }
TextEditorCodeExecutionToolResultError object{ type: "text_editor_code_execution_tool_result_error", error_code, error_message }
type: "text_editor_code_execution_tool_result_error"
error_code: TextEditorCodeExecutionToolResultErrorCode
TextEditorCodeExecutionToolResultErrorCode = "invalid_tool_input" or "unavailable" or "too_many_requests" or 2 more
TextEditorCodeExecutionToolResultErrorParam object{ type: "text_editor_code_execution_tool_result_error", error_code, error_message }
error_code: TextEditorCodeExecutionToolResultErrorCode
TextEditorCodeExecutionViewResultBlock object{ type: "text_editor_code_execution_view_result", content, file_type, 3 more }
type: "text_editor_code_execution_view_result"
file_type: "text" or "image" or "pdf"
TextEditorCodeExecutionViewResultBlockParam object{ type: "text_editor_code_execution_view_result", content, file_type, 3 more }
file_type: "text" or "image" or "pdf"
ThinkingBlock object{ type: "thinking", signature, thinking }
type: "thinking"
signature: stringA value used to verify that this thinking block was generated by Claude when it is passed back to the API.
A value used to verify that this thinking block was generated by Claude when it is passed back to the API.
This is an opaque field and should not be interpreted or parsed. When passing thinking blocks back to the API (required when using tools with extended thinking), pass them back exactly as received, with this field intact.
See extended thinking for details.
The text of Claude's thinking process for this block.
ThinkingBlockParam object{ type: "thinking", signature, thinking }
signature: stringThe signature value of this thinking block, exactly as returned by the API in a previous response. Used to verify that the block was generated by Claude.
The signature value of this thinking block, exactly as returned by the API in a previous response. Used to verify that the block was generated by Claude.
Thinking blocks must be passed back unmodified and in their original order; a modified block results in a 400 invalid_request_error.
The thinking text of this block as returned by the API.
ThinkingConfigAdaptive object{ type: "adaptive", display }
display: optional "summarized" or "omitted" or nullControls how thinking content appears in the response. When set to summarized, thinking is returned normally. When set to omitted, thinking content is redacted but a signature is returned for multi-turn continuity. The default depends on the model; see Controlling thinking display.
Controls how thinking content appears in the response. When set to summarized, thinking is returned normally. When set to omitted, thinking content is redacted but a signature is returned for multi-turn continuity. The default depends on the model; see Controlling thinking display.
ThinkingConfigBetweenTools object{ type: "between_tools" }
ThinkingConfigDisabled object{ type: "disabled" }
ThinkingConfigEnabled object{ type: "enabled", budget_tokens, display }
budget_tokens: numberDetermines how many tokens Claude can use for its internal reasoning process. Larger budgets can enable more thorough analysis for complex problems, improving response quality.
Determines how many tokens Claude can use for its internal reasoning process. Larger budgets can enable more thorough analysis for complex problems, improving response quality.
Must be ≥1024 and less than max_tokens.
See extended thinking for details.
display: optional "summarized" or "omitted" or nullControls how thinking content appears in the response. When set to summarized, thinking is returned normally. When set to omitted, thinking content is redacted but a signature is returned for multi-turn continuity. The default depends on the model; see Controlling thinking display.
Controls how thinking content appears in the response. When set to summarized, thinking is returned normally. When set to omitted, thinking content is redacted but a signature is returned for multi-turn continuity. The default depends on the model; see Controlling thinking display.
ThinkingConfigParam = ThinkingConfigEnabled or ThinkingConfigDisabled or ThinkingConfigBetweenTools or ThinkingConfigAdaptiveConfiguration for Claude's thinking.
Configuration for Claude's thinking.
With {"type": "adaptive"}, Claude decides when and how much to think. With {"type": "enabled"} (manual extended thinking), you set a budget_tokens of at least 1,024. Thinking tokens count toward your max_tokens limit.
Which type values are accepted, and what happens when you omit thinking, depend on the model. See thinking for each model's behavior.
ThinkingDelta object{ type: "thinking_delta", thinking }
type: "thinking_delta"
The incremental thinking text for this content block. Concatenate the thinking values of successive thinking_delta events to assemble the block's full thinking value.
Tool object{ type, input_schema, name, 7 more }
ToolBash20250124 object{ type: "bash_20250124", name, allowed_callers, 4 more }
ToolChoice = ToolChoiceAuto or ToolChoiceAny or ToolChoiceTool or ToolChoiceNoneHow the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all.
How the model should use the provided tools. The model can use a specific tool, any available tool, decide by itself, or not use tools at all.
ToolChoiceAny object{ type: "any", disable_parallel_tool_use }The model will use any available tools.
The model will use any available tools.
disable_parallel_tool_use: optional booleanWhether to disable parallel tool use.
Whether to disable parallel tool use.
Defaults to false. If set to true, the model will output exactly one tool use.
ToolChoiceAuto object{ type: "auto", disable_parallel_tool_use }The model will automatically decide whether to use tools.
The model will automatically decide whether to use tools.
disable_parallel_tool_use: optional booleanWhether to disable parallel tool use.
Whether to disable parallel tool use.
Defaults to false. If set to true, the model will output at most one tool use.
ToolChoiceNone object{ type: "none" }The model will not be allowed to use tools.
The model will not be allowed to use tools.
ToolChoiceTool object{ type: "tool", name, disable_parallel_tool_use }The model will use the specified tool with tool_choice.name.
The model will use the specified tool with tool_choice.name.
The name of the tool to use.
disable_parallel_tool_use: optional booleanWhether to disable parallel tool use.
Whether to disable parallel tool use.
Defaults to false. If set to true, the model will output exactly one tool use.
ToolReferenceBlock object{ type: "tool_reference", tool_name }
type: "tool_reference"
tool_name: string
ToolReferenceBlockParam object{ type: "tool_reference", tool_name, cache_control }Tool reference block that can be included in tool_result content.
Tool reference block that can be included in tool_result content.
tool_name: string
cache_control: optional CacheControlEphemeral { type: "ephemeral", ttl } or nullCreate a cache control breakpoint at this content block.
Create a cache control breakpoint at this content block.
ttl: optional "5m" or "1h"The time-to-live for the cache control breakpoint.
The time-to-live for the cache control breakpoint.
This may be one the following values:
5m: 5 minutes1h: 1 hour
Defaults to 5m. See prompt caching pricing for details.
ToolResultBlockParam object{ type: "tool_result", tool_use_id, cache_control, 3 more }
ToolSearchToolBm25_20251119 object{ type, name, allowed_callers, 3 more }
ToolSearchToolRegex20251119 object{ type, name, allowed_callers, 3 more }
ToolSearchToolResultBlock object{ type: "tool_search_tool_result", content, tool_use_id }
ToolSearchToolResultBlockParam object{ type: "tool_search_tool_result", content, tool_use_id, cache_control }
ToolSearchToolResultError object{ type: "tool_search_tool_result_error", error_code, error_message }
type: "tool_search_tool_result_error"
error_code: ToolSearchToolResultErrorCode
ToolSearchToolResultErrorCode = "invalid_tool_input" or "unavailable" or "too_many_requests" or "execution_time_exceeded"
ToolSearchToolResultErrorParam object{ type: "tool_search_tool_result_error", error_code, error_message }
error_code: ToolSearchToolResultErrorCode
ToolSearchToolSearchResultBlock object{ type: "tool_search_tool_search_result", tool_references }
type: "tool_search_tool_search_result"
tool_references: array of ToolReferenceBlock { type: "tool_reference", tool_name }
type: "tool_reference"
tool_name: string
ToolSearchToolSearchResultBlockParam object{ type: "tool_search_tool_search_result", tool_references }
ToolTextEditor20250124 object{ type: "text_editor_20250124", name, allowed_callers, 4 more }
ToolTextEditor20250429 object{ type: "text_editor_20250429", name, allowed_callers, 4 more }
ToolTextEditor20250728 object{ type: "text_editor_20250728", name, allowed_callers, 5 more }
ToolUnion = Tool or ToolBash20250124 or CodeExecutionTool20250522 or 18 more
ToolUseBlock object{ type: "tool_use", id, caller, 3 more }
ToolUseBlockParam object{ type: "tool_use", id, input, 4 more }
ToolUseCaller = DirectCaller or ServerToolCaller or ServerToolCaller20260120Which party invoked the tool call: the model directly, or a server tool on its behalf.
Which party invoked the tool call: the model directly, or a server tool on its behalf.
DirectCaller object{ type: "direct" }Tool invocation directly from the model.
Tool invocation directly from the model.
ServerToolCaller object{ type: "code_execution_20250825", tool_id }Tool invocation generated by a server-side tool.
Tool invocation generated by a server-side tool.
tool_id: string
ServerToolCaller20260120 object{ type: "code_execution_20260120", tool_id }
tool_id: string
ToolsetToolUseBlock = BrowserToolUseBlock or ComputerToolUseBlock or ToolUseBlock
URLImageSource object{ type: "url", url }
URLPDFSource object{ type: "url", url }
Usage object{ cache_creation, cache_creation_input_tokens, cache_read_input_tokens, 6 more }
UserLocation object{ type: "approximate", city, country, 2 more }
city: optional string or nullThe city of the user.
The city of the user.
country: optional string or nullThe two letter ISO country code of the user.
The two letter ISO country code of the user.
region: optional string or nullThe region of the user.
The region of the user.
timezone: optional string or nullThe IANA timezone of the user.
The IANA timezone of the user.
WebFetchBlock object{ type: "web_fetch_result", content, retrieved_at, url }
type: "web_fetch_result"
content: DocumentBlock { type: "document", citations, source, title }
ISO 8601 timestamp when the content was retrieved
Fetched content URL
WebFetchBlockParam object{ type: "web_fetch_result", content, url, retrieved_at }
content: DocumentBlockParam { type: "document", source, cache_control, 3 more }
Fetched content URL
ISO 8601 timestamp when the content was retrieved
WebFetchTool20250910 object{ type: "web_fetch_20250910", name, allowed_callers, 9 more }
WebFetchTool20260209 object{ type: "web_fetch_20260209", name, allowed_callers, 9 more }
WebFetchTool20260309 object{ type: "web_fetch_20260309", name, allowed_callers, 10 more }Web fetch tool with use_cache parameter for bypassing cached content.
Web fetch tool with use_cache parameter for bypassing cached content.
WebFetchTool20260318 object{ type: "web_fetch_20260318", name, allowed_callers, 11 more }
WebFetchToolResultBlock object{ type: "web_fetch_tool_result", caller, content, tool_use_id }
WebFetchToolResultBlockParam object{ type: "web_fetch_tool_result", content, tool_use_id, 2 more }
WebFetchToolResultErrorBlock object{ type: "web_fetch_tool_result_error", error_code }
type: "web_fetch_tool_result_error"
error_code: WebFetchToolResultErrorCode
WebFetchToolResultErrorBlockParam object{ type: "web_fetch_tool_result_error", error_code }
error_code: WebFetchToolResultErrorCode
WebFetchToolResultErrorCode = "invalid_tool_input" or "url_too_long" or "url_not_allowed" or 7 more
WebFetchURLSourceAll object{ type: "all" }The url_sources variant under which a source contributes in
full: every result of the tool filter's source, or all user input.
The url_sources variant under which a source contributes in
full: every result of the tool filter's source, or all user input.
WebFetchURLSourceExcept object{ type: "except", tools }The tool filter variant under which every result but the named
tools' contributes.
The tool filter variant under which every result but the named tools' contributes.
tools: array of WebFetchURLSourceToolReference { type: "tool_reference", name }
WebFetchURLSourceNone object{ type: "none" }The url_sources variant under which a source contributes nothing:
no result of the tool filter's source, or no user input.
The url_sources variant under which a source contributes nothing:
no result of the tool filter's source, or no user input.
WebFetchURLSourceOnly object{ type: "only", tools }The tool filter variant under which only the named tools' results
contribute.
The tool filter variant under which only the named tools' results contribute.
tools: array of WebFetchURLSourceToolReference { type: "tool_reference", name }
WebFetchURLSourceToolReference object{ type: "tool_reference", name }One entry of a tool filter's tools: it must name a tool declared
in this request's tools[].
One entry of a tool filter's tools: it must name a tool declared
in this request's tools[].
WebFetchURLSources object{ client_tool_results, server_tool_results, user_input }Which sources contribute to the set of URLs web fetch may fetch.
Which sources contribute to the set of URLs web fetch may fetch.
Each key is a tagged variant: user_input is all or none; the
two tool filters are all, none, only (only the named tools'
results) or except (every result but the named tools'). A named tool
must be declared in this request's tools[].
WebSearchResultBlock object{ type: "web_search_result", encrypted_content, page_age, 2 more }
type: "web_search_result"
WebSearchResultBlockParam object{ type: "web_search_result", encrypted_content, title, 2 more }
WebSearchTool20250305 object{ type: "web_search_20250305", name, allowed_callers, 7 more }
WebSearchTool20260209 object{ type: "web_search_20260209", name, allowed_callers, 7 more }
WebSearchTool20260318 object{ type: "web_search_20260318", name, allowed_callers, 8 more }
WebSearchToolRequestError object{ type: "web_search_tool_result_error", error_code }
error_code: WebSearchToolResultErrorCode
WebSearchToolResultBlock object{ type: "web_search_tool_result", caller, content, tool_use_id }
WebSearchToolResultBlockContent = WebSearchToolResultError or array of WebSearchResultBlock
WebSearchToolResultBlockParam object{ type: "web_search_tool_result", content, tool_use_id, 2 more }
WebSearchToolResultBlockParamContent = array of WebSearchResultBlockParam or WebSearchToolRequestError
WebSearchToolResultError object{ type: "web_search_tool_result_error", error_code }
type: "web_search_tool_result_error"
error_code: WebSearchToolResultErrorCode
WebSearchToolResultErrorCode = "invalid_tool_input" or "unavailable" or "max_uses_exceeded" or 3 more
MessagesBatches
Retrieve a Message Batch
This endpoint is idempotent and can be used to poll for Message Batch completion. To access the results of a Message Batch, make a request to the results_url field in the response.
List Message Batches
List all Message Batches within a Workspace. Most recently created batches are returned first.
Cancel a Message Batch
Batches may be canceled any time before processing ends. Once cancellation is initiated, the batch enters a canceling state, at which time the system may complete any in-progress, non-interruptible requests before finalizing cancellation.
Delete a Message Batch
Retrieve Message Batch results
Streams the results of a Message Batch as a .jsonl file.