Struct MemoryConfig 

pub struct MemoryConfig {

    pub backend: MemoryBackendKind,
    pub workspace: String,
    pub embedding_provider: String,
    pub embedding_model: String,
    pub gemini_api_key: Option<String>,
    pub index_sessions: bool,
    pub multimodal_embeddings: bool,
    pub embedding_cache_dir: String,
    pub chunk_size: usize,
    pub chunk_overlap: usize,
    pub paths: Vec<MemoryIndexPath>,
    pub session_max_messages: usize,
    pub session_max_chars: usize,
    pub temporal_decay_lambda: f64,
    pub llm_query_expansion: bool,
    pub wiki_enabled: bool,
    pub wiki_fresh_days: u32,
    pub wiki_stale_days: u32,
}

Fields

backend: MemoryBackendKind

Memory search backend: “sqlite” (default), “markdown”, or “none”

workspace: String

embedding_provider: String

Embedding provider: “local” (fastembed, default), “openai”, “gemini”, or “none”

embedding_model: String

gemini_api_key: Option<String>

Gemini API key for embeddings (supports ${ENV_VAR} syntax)

index_sessions: bool

Index completed session transcripts for memory_search (default: false)

multimodal_embeddings: bool

Enable multimodal embeddings for images (requires gemini-embedding-2-preview model)

embedding_cache_dir: String

Cache directory for local embedding models (optional) Default: ~/.cache/localgpt/models Can also be set via FASTEMBED_CACHE_DIR environment variable

chunk_size: usize

chunk_overlap: usize

paths: Vec<MemoryIndexPath>

Additional paths to index (relative to workspace or absolute) Each path uses a glob pattern for file matching

session_max_messages: usize

Maximum messages to save in session memory files (0 = unlimited) Similar to OpenClaw’s hooks.session-memory.messages (default: 15)

session_max_chars: usize

Maximum characters per message in session memory (0 = unlimited) Set to 0 to preserve full message content like OpenClaw

temporal_decay_lambda: f64

Temporal decay factor for search scoring. Older memories get lower scores using: score * exp(-lambda * age_days) Default: 0.0 (disabled) 0.1 = ~50% penalty for 7-day old memory 0.05 = ~50% penalty for 14-day old memory

llm_query_expansion: bool

Enable LLM-based query expansion for memory search. Costs one extra LLM call per search but produces better keywords from conversational queries. Default: false.

wiki_enabled: bool

Enable the Memory Wiki structured knowledge layer (claims, evidence, staleness). Default: false.

wiki_fresh_days: u32

Days after last update before a wiki claim transitions from Fresh to Aging. Default: 30.

wiki_stale_days: u32

Days after last update before a wiki claim transitions from Aging to Stale. Default: 90.

Trait Implementations

impl Clone for MemoryConfig

fn clone(&self) -> MemoryConfig

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for MemoryConfig

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl Default for MemoryConfig

fn default() -> Self

Returns the “default value” for a type.

impl<'de> Deserialize<'de> for MemoryConfig

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for MemoryConfig

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.