Class MemoryDeleteTool

ITool implementation that soft-deletes a memory trace from the agent's SQLite brain database.

Usage:

const tool = new MemoryDeleteTool(brain);
const result = await tool.execute(
  { traceId: 'mt_3f2b8c1e-5d4a-4e6f-9b7c-2a1d0e8f6c5b', reason: 'Information is outdated.' },
  context,
);
// result.output.deleted → true

Implements

  • ITool<MemoryDeleteInput, MemoryDeleteOutput>

Constructors

constructor

Methods

refusalBeforeApproval execute

Properties

id name displayName description category hasSideEffects inputSchema

Constructors

constructor

  • new MemoryDeleteTool(brain, writeBack?): MemoryDeleteTool

  • Parameters

    • brain: Brain

      The agent's shared SQLite brain database connection.

    • Optional writeBack: MemoryToolWriteBack

      The facade that made the tool (Memory.createTools()): a deleted trace's knowledge-graph copy is deleted through it, and a delete a learning layer does not keep is taken back through it. Without it the tool restores the row itself.

    Returns MemoryDeleteTool

Methods

refusalBeforeApproval

  • refusalBeforeApproval(_args, context): undefined | ToolExecutionResult<MemoryDeleteOutput>

  • The refusal a delete meets whatever a human would answer, read before the tool orchestrator's approval gate (ITool.refusalBeforeApproval), so no one is asked to approve a write that cannot act: a freeze, and, with the turn's scope ids on the call, a session that records no writes. A refusal that reads the trace (other_user, conflict) is given when the call runs. execute refuses the same call itself, in the same order.

    Parameters

    • _args: Record<string, any>

      The call's arguments, before the input schema is checked.

    • context: ToolExecutionContext

      The call's context.

    Returns undefined | ToolExecutionResult<MemoryDeleteOutput>

    The refusal, or undefined when the call may run.

execute

  • execute(args, context): Promise<ToolExecutionResult<MemoryDeleteOutput>>

  • Set deleted = 1 on the memory trace matching traceId.

    The WHERE clause filters to deleted = 0 so that attempting to delete an already-deleted trace returns { deleted: false } rather than silently succeeding — this gives callers accurate feedback.

    The trace's knowledge-graph copy (the episodic record a memory manager on the same Brain writes of each trace it stores) is deleted with it, through the facade that made the tool.

    Parameters

    • args: MemoryDeleteInput

      Delete input (traceId, optional reason).

    • context: ToolExecutionContext

      Tool execution context; where it carries the turn's scope ids, the call reaches only a live trace of the turn's own scopes that the turn's own user filed (another scope's reads as not found, another user's is refused, other_user), and the delete runs through the learner session's recorder, deleting only while the trace holds what was read (conflict otherwise). Only a freeze stops a delete, with a layer or without one (frozen).

    Returns Promise<ToolExecutionResult<MemoryDeleteOutput>>

    { deleted } status, or an error result.

Properties

Readonly id

id: "memory-delete-v1" = 'memory-delete-v1'

Globally unique tool identifier.

Readonly name

name: "memory_delete" = 'memory_delete'

LLM-facing tool name.

Readonly displayName

displayName: "Delete Memory" = 'Delete Memory'

Human-readable display name.

Readonly description

description: string = ...

LLM-facing description.

Readonly category

category: "memory" = 'memory'

Logical category for discovery and grouping.

Readonly hasSideEffects

hasSideEffects: true = true

This tool writes to the database.

Readonly inputSchema

inputSchema: JSONSchemaObject = ...

JSON schema for input validation and LLM tool-call construction.

Settings

Member Visibility

Theme

On This Page

constructorrefusalBeforeApprovalexecuteidnamedisplayNamedescriptioncategoryhasSideEffectsinputSchema