Struct MemoryMetadata 

pub struct MemoryMetadata {

    pub id: MemoryId,
    pub galaxy: Galaxy,
    pub content_hash: String,
    pub tags: Vec<String>,
    pub importance: f32,
    pub created_at: DateTime<Utc>,
    pub accessed_at: DateTime<Utc>,
    pub access_count: u64,
    pub coords: HolographicCoords,
    pub coord5d: Coordinate5D,
    pub memory_type: MemoryType,
    pub neuro_score: f32,
    pub novelty_score: f32,
    pub emotional_valence: f32,
    pub emotional_weight: f32,
    pub is_protected: bool,
    pub is_private: bool,
    pub model_exclude: bool,
    pub source: String,
    pub source_trust: f32,
    pub half_life_days: f32,
    pub recall_count: u64,
    pub version: u64,
    pub agent_id: String,
    pub title: Option<String>,
    pub topic: Option<String>,
    pub tier: Tier,
    pub class: Option<MemoryClass>,
    pub dup_count: u64,
    pub validity: ValidityState,
    pub corroborated_by: Vec<Uuid>,
    pub revision_count: u32,
}

Metadata for a memory entry.

Fields

id: MemoryId

Memory UUID

galaxy: Galaxy

Which galaxy this memory lives in

content_hash: String

Content hash for deduplication

tags: Vec<String>

Tags for categorization

importance: f32

Importance score (0.0 to 1.0)

created_at: DateTime<Utc>

Creation timestamp

accessed_at: DateTime<Utc>

Last accessed timestamp

access_count: u64

Access count

coords: HolographicCoords

Holographic coordinates

coord5d: Coordinate5D

5D holographic coordinate for spatial indexing

memory_type: MemoryType

Memory classification

neuro_score: f32

Dynamic neural strength (0.0-1.0) — decays over time, boosts on recall

novelty_score: f32

Novelty score (0.0-1.0) — decays as info becomes familiar

emotional_valence: f32

Emotional valence (-1.0 = negative, 1.0 = positive)

emotional_weight: f32

Emotional weight / resonance (0.0-1.0)

is_protected: bool

Hard protection from forgetting

is_private: bool

Exclude from MCP tool responses

model_exclude: bool

Exclude from AI model context windows

source: String

Provenance: “user”, “tool”, “inferred”, “web”

source_trust: f32

Trust score for source (0.0-1.0, defends against memory poisoning)

half_life_days: f32

Per-memory configurable decay half-life in days

recall_count: u64

Recall count (independent from access_count)

version: u64

Version for multi-agent cache coherence

agent_id: String

Last writer identity

title: Option<String>

Human-readable title (envelope v2, S4). None = untitled; absent in old records deserializes as None, never as a fabricated value.

topic: Option<String>

Topic label for subject-scoped retrieval (envelope v2, S4)

tier: Tier

Lifecycle tier (V8 S5) — stamped at creation, dream-cycle-only transitions. Pre-S5 rows default to Tier::Episodic.

class: Option<MemoryClass>

Typology class (V8 S5, crate::typology) — stamped at creation when confidently recognized. None = unstamped (pre-S5 row, or content the detector does not confidently recognize); never fabricated as a claim about the content.

dup_count: u64

Duplicate-write counter (V8 S5 dedup gate) — bumped instead of re-inserting identical content; importance decays with it.

validity: ValidityState

Lifecycle validity (V8 Slice B, D1+D2) — reuses wm_core::episodic::ValidityState; the dream cycle is the only transition path (see validity_sweep). Pre-Slice-B rows default to Active, so the recall surface is byte-identical until the WM_VALIDITY_ENFORCE knob turns on.

corroborated_by: Vec<Uuid>

Corroborating session ids (bridging consensus counter, 2nd PR after Slice B) — distinct sessions whose agents independently stand behind this memory, recorded via memory.corroborate. Empty = uncorroborated (the pre-counter default for every row). Sessions nest agents and machines (maintainer ruling 2026-09-04: sessions are the working unit for now); each entry is client-asserted like user_id — attribution, never authentication.

revision_count: u32

Content-revision counter (V8 S11c) — how many times this memory’s content has changed through memory.update. The per-entry chain itself lives in the store’s revisions DBI (memory.revisions).

Trait Implementations

impl Clone for MemoryMetadata

fn clone(&self) -> MemoryMetadata

Returns a duplicate of the value.

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

Performs copy-assignment from source.

impl Debug for MemoryMetadata

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

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for MemoryMetadata

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

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for MemoryMetadata

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

where __S: Serializer,

Serialize this value into the given Serde serializer.