kael-bit/engram-rs

AI Agents & Assistants

A Rust‑based memory engine for AI agents that adds forgetting, promotion, and self‑organizing topic clustering to flat stores

Rust Latest v0.17.1 · 3mo ago Security brief →

Features

Three‑layer lifecycle (Buffer → Working → Core) with decay and LLM quality gating
Automatic semantic deduplication and merging of similar memories
Self‑organizing topic tree that clusters related knowledge without manual tags

View all 16 releases →

No immediate action

v0.17.0 New feature 3mo

Incremental insert + naming improvements

Open

Review required

v0.16.1 Breaking risk 3mo

Auth Dependencies

Data loss & indexing fixes

Open

No immediate action

v0.16.0 Breaking risk 3mo

IDF-weighted recall overhaul

Open

No immediate action

v0.15.0 Breaking risk 3mo

Multi-agent support

Open

v0.14.0 Breaking risk 3mo

Breaking changes

`rerank` field removed from `RecallRequest`
Global audit replaced with focused topic distillation (condenses overlapping entries when a topic reaches 10+ memories)
Sandbox functionality removed

Notable features

Episodic Core block: episodic memories cannot be promoted to Core (enforced at collection and DB level)
Web UI split into `index.html`, `style.css`, and `app.js` for easier maintenance
Sort by weight added to web UI memory list with four modes (Newest, Oldest, Weight ↓, Weight ↑)

Full changelog

Rerank removed — rerank field removed from RecallRequest. LLM re-ranking was default-off and unnecessary; FTS + semantic scoring is sufficient. Reduces recall latency.
Audit → Topic Distillation — old three-powers global audit replaced with focused topic distillation. When a topic grows to 10+ memories, overlapping entries are condensed into fewer, richer ones. More efficient, fewer wasted LLM calls.
Sandbox removed — no longer needed with the simpler distillation model.

Episodic Core block — episodic memories can never be promoted to Core. Enforced at both candidate collection and DB level.
Web UI split — single 1300-line index.html split into index.html + style.css + app.js. Easier to maintain and read.
Sort by weight — web UI memory list supports 4 sort modes: Newest, Oldest, Weight ↓, Weight ↑. Uses the same memory_weight() formula as recall scoring.

Centralized prompts — all LLM prompts and JSON tool schemas in src/prompts.rs
Centralized thresholds — all 60+ magic numbers from 15+ files in src/thresholds.rs
Epoch-based decay — decay only during active consolidation, not wall-clock time
Buffer redesign — all new memories enter Buffer; promotion requires consolidation
Reconcile merge — auto-detects same-topic memories where newer supersedes older
~1400 lines removed net across refactoring and feature removal