Crate recall_echo 

recall-echo — Persistent memory system with knowledge graph.

A general-purpose persistent memory system for any LLM tool — Claude Code, Ollama, or any provider. Features a four-layer memory architecture with a knowledge graph (SurrealDB + fastembed) as Layer 0.

Architecture

Input adapters (JSONL transcripts, pulse-null Messages)
    → Conversation (universal internal format)
    → Archive pipeline (markdown + index + ephemeral + graph)

Features

• graph (default) — Knowledge graph with SurrealDB + fastembed
• pulse-null — Plugin integration for pulse-null entities

Re-exports

pub use archive::SessionMetadata;

pub use summarize::ConversationSummary;

Modules

archive

Conversation archival — converts conversations into persistent markdown archives.

checkpoint

Checkpoint creation — saves a conversation snapshot before context compression.

config

config_cli

CLI handlers for recall-echo config show and recall-echo config set.

consume

EPHEMERAL.md consumption — reads recent session context.

conversation

Core conversation types and processing.

dashboard

distill

ephemeral

frontmatter

graph

recall-graph — Knowledge graph with semantic search for AI memory systems.

graph_bridge

Bridge between recall-echo and recall-graph.

graph_cli

Graph memory CLI subcommands (behind graph feature flag).

init

Initialize the recall-echo memory system.

jsonl

JSONL transcript parsing for Claude Code sessions.

paths

Path resolution utilities for recall-echo.

search

status

Memory system health check.

summarize

Conversation summarization with optional LLM enhancement.

tags

Structured tag extraction from conversations.

Structs

RecallEcho

The recall-echo memory system.