chooser · Ue MCP: Risk & Policy

WHAT IT DOES

What chooser does on Ue

AI agents use chooser to create or update resources in Ue — usually the action step of a workflow, after the agent has gathered context. Every call changes real data in your Ue environment.

PARAMETERS

Parameter	Type	Required	Description
`name`	string	—	create: new ChooserTable asset name
`cells`	array	—	Per-column cell values aligned to column order; each is struct text like '(Value=2)', or a scalar. null/omitted leaves a column at its default.
`index`	number	—	Row index for set_row / delete_row
`table`	string	—	ChooserTable asset path, e.g. /Game/Path/CT_Locomotion
`action`	string	Yes	Action to perform
`inputs`	object	—	Cell values keyed by column index (as string) or column name; same value format as cells
`output`	string	—	Output asset path for the row (a PoseSearchDatabase, a nested ChooserTable, etc.)
`disabled`	boolean	—	set_row: enable/disable the row without deleting it
`enumPath`	string	—	add_column: enum asset path for an EnumColumn
`columnType`	string	—	add_column: Chooser column struct short name (EnumColumn, BoolColumn, FloatRangeColumn, GameplayTagColumn, ObjectColumn, Output*Column, ...)
`onConflict`	string	—	create: conflict policy skip (default) \| error \| overwrite
`outputType`	string	—	Output wrapper: 'asset' (hard ref, default) \| 'soft_asset' \| 'evaluate' (nested ChooserTable reference)

Parameters from the server's own tool schema.

WHY IT NEEDS A POLICY

Why chooser needs a policy

The tool creates and modifies ChooserTable assets (add/edit/delete rows and columns), which are reversible data modifications within the Unreal Engine editor. The most severe action here is deletion of rows, but since these are editor assets that can be recreated or undone within the editor context, this falls under Write rather than Destructive.

From the tool's definition Author ChooserTable assets... introspect columns, list/add/edit/delete rows mapping input-column conditions to an output object; actions include create, add_column, add_row, edit, delete rows

Risk signalsHigh parameter count (15 properties) · Bulk/mass operation — affects multiple targets

Attacks that exploit this kind of access

Questions about chooser

What does the chooser tool do? +

Author ChooserTable assets (the data-driven selection layer behind Motion Matching): introspect columns, list/add/edit/delete rows mapping input-column conditions to an output object. Actions: - create: Create an empty ChooserTable asset. Add input columns with add_column, then rows with add_row. Params: name, packagePath? (default /Game), onConflict? (#685) - describe: Introspect a ChooserTable: row count, each input column (index, name, columnType, cellType) and the fallback result. Read this first to learn the cell text format each column expects. Params: table (#685) - add_column: Add an input column to a ChooserTable (so rows have a condition to fill). columnType is a Chooser column struct short name, e.g. EnumColumn, BoolColumn, FloatRangeColumn, GameplayTagColumn, ObjectColumn, or an Output* column. Optionally bind its input: inputStruct (parameter struct e.g. EnumContextProperty/BoolContextProperty), boundProperty (context property name to read), enumPath (for enum columns). Sizes the new column's cells to the current rows. Params: table, columnType, inputStruct?, boundProperty?, enumPath? (#685) - list_rows: List every row: index, disabled flag, output object (resultType + referenced asset path), and each column's cell value as round-trippable text. Params: table (#685) - add_row: Append a row. Set the output via output (asset path) + outputType ('asset' hard ref default | 'soft_asset' | 'evaluate' for a nested ChooserTable). Set input-column conditions via cells (array aligned to column order) and/or inputs (object keyed by column index or name). Cell values are struct text like '(Value=2)' - partial fields are allowed and unspecified ones keep defaults; a bare number/bool works for scalar columns. Params: table, output?, outputType?, cells?, inputs? (#685) - set_row: Edit an existing row by index: optionally replace the output (output + outputType), toggle disabled, and/or update column cells (cells / inputs, same format as add_row). Params: table, index, output?, outputType?, disabled?, cells?, inputs? (#685) - delete_row: Delete a row by index (removes its output plus the per-row cell from every column). Params: table, index (#685). It is categorised as a Write tool in the Ue MCP Server, which means it can create or modify data. Consider rate limits to prevent runaway writes.

What parameters does chooser accept? +

chooser accepts 12 parameters: name, cells, index, table, action, inputs, output, disabled, enumPath, columnType, onConflict, outputType. Required: action. The full parameter table on this page comes from the server's own tool schema.

How do I enforce a policy on chooser? +

Register the Ue MCP server in PolicyLayer and add a rule for chooser: allow, deny, rate-limit, or require approval. Point your MCP client at the PolicyLayer proxy URL and the rule is enforced on every call, before it reaches Ue. Nothing to install.

What risk level is chooser? +

chooser is a Write tool with medium risk. Write tools should be rate-limited to prevent accidental bulk modifications.

Can I rate-limit chooser? +

Yes. Add a rate_limit block to the chooser rule in your PolicyLayer policy. For example, setting max: 10 and window: 60 limits the tool to 10 calls per minute. Rate limits are tracked per agent session and reset automatically.

How do I block chooser completely? +

Set action: deny in the PolicyLayer policy for chooser. The AI agent will receive a policy violation error and cannot call the tool. You can also include a reason field to explain why the tool is blocked.

What MCP server provides chooser? +

chooser is provided by the Ue MCP server (ue-mcp). PolicyLayer sits as a proxy in front of this server to enforce policies before tool calls reach the server.