create_context_management_config: Create Context Management Configuration

In aisdk: Unified Interface for AI Model Providers

Create Context Management Configuration

Description

Build a normalized context-management configuration list.

Usage

create_context_management_config(
  mode = NULL,
  llm_synthesis = FALSE,
  synthesis_model = NULL,
  llm_synthesis_policy = NULL,
  context_window_override = NULL,
  max_output_tokens_override = NULL,
  project_memory_root = NULL,
  retrieval_providers = NULL,
  retrieval_provider_order = NULL,
  retrieval_provider_limits = NULL,
  retrieval_min_hits = NULL,
  retrieval_scoring_policy = NULL,
  retrieval_reranking = FALSE,
  retrieval_reranking_model = NULL,
  retrieval_reranking_policy = NULL
)

Arguments

mode	One of "off", "basic", or "adaptive".
llm_synthesis	Logical; whether to enable optional LLM enrichment for working-memory synthesis.
synthesis_model	Optional model reference used for LLM synthesis.
llm_synthesis_policy	Optional named list controlling LLM synthesis behavior. Supported fields are min_regime, recent_messages, max_items, and temperature.
context_window_override	Optional numeric context-window override.
max_output_tokens_override	Optional numeric max-output override.
project_memory_root	Optional project root used for ProjectMemory retrieval and learning.
retrieval_providers	Optional retrieval-provider selection. Accepts either a character vector of enabled providers or a named logical/list map.
retrieval_provider_order	Optional character vector controlling render and ranking priority across retrieval providers.
retrieval_provider_limits	Optional named list of per-provider result limits.
retrieval_min_hits	Optional named list of per-provider matching thresholds.
retrieval_scoring_policy	Optional named list controlling cross-provider retrieval scoring. Supported fields are provider_weights, token_match_weight, exact_query_bonus, title_match_weight, summary_match_weight, preview_match_weight, bigram_match_weight, coverage_weight, workflow_hint_weight, accessor_match_weight, semantic_object_bonus, task_signal_weight, execution_error_weight, system_metric_weight, runtime_signal_weight, source_kind_bonus, recency_weight, and max_total_results.
retrieval_reranking	Logical; whether to enable optional learned reranking on top of deterministic retrieval scoring.
retrieval_reranking_model	Optional model reference used for reranking.
retrieval_reranking_policy	Optional named list controlling reranking. Supported fields are min_regime, top_n, and temperature.

Value

A normalized configuration list.

aisdk documentation built on May 29, 2026, 9:07 a.m.