import type { MemoryCitationsMode } from "../config/types.memory.js"; import type { ContextEngine, CompactResult } from "./types.js"; /** * Delegate a context-engine compaction request to OpenClaw's built-in runtime compaction path. * * This is the same bridge used by the legacy context engine. Third-party * engines can call it from their own `compact()` implementations when they do * not own the compaction algorithm but still need `/compact` and overflow * recovery to use the stock runtime behavior. * * Note: `compactionTarget` is part of the public `compact()` contract, but the * built-in runtime compaction path does not expose that knob. This helper * ignores it to preserve legacy behavior; engines that need target-specific * compaction should implement their own `compact()` algorithm. */ export declare function delegateCompactionToRuntime(params: Parameters[0]): Promise; /** * Build a context-engine-ready systemPromptAddition from the active memory * plugin prompt path. This lets non-legacy engines explicitly opt into the * same memory/wiki guidance that the legacy engine gets via system prompt * assembly, without reimplementing memory prompt formatting. */ export declare function buildMemorySystemPromptAddition(params: { availableTools: Set; citationsMode?: MemoryCitationsMode; }): string | undefined;