Hermes Atlas
Memory & knowledge

Hermes Ladybug Memory

Ladybug-Memory/hermes-memory-plugin

Local memory provider for Hermes Agent backed by an embedded Ladybug graph database

In short

Hermes Ladybug Memory is a memory-provider plugin that stores Hermes Agent memory in a local Ladybug Memory graph database file. It gives Hermes Agent users fully local, searchable memory with typed entries and graph links, and needs no API keys or cloud service.

What Hermes Ladybug Memory does

The plugin keeps everything in one .lbdb file in HERMES_HOME, backed by LadybugMemory, a columnar embedded graph database. It supports BM25 keyword search, importance-weighted recall, typed memory entries, named graph edges between entries and optional GLiNER2 entity extraction. Before every turn it prefetches memories in the background, six by default, using importance-weighted recall plus a query search.

The agent gets eight tools: ladybug_store, ladybug_search, ladybug_recall, ladybug_update, ladybug_delete, ladybug_link, ladybug_related and ladybug_entity. The plugin also mirrors writes to the built-in MEMORY.md and USER.md into Ladybug and surfaces high-importance memories during context compression. Settings live under memory.ladybug in config.yaml: db_path, prefetch_limit, min_importance and auto_link. The README positions it against Honcho, Mem0 and RetainDB as the option with no hosted component.

Key features

  • Single-file local .lbdb store with no API keys, cloud or sync service
  • BM25 keyword search and importance-weighted recall
  • Typed memory entries and named graph links between memories
  • Optional GLiNER2 entity extraction
  • Mirrors built-in MEMORY.md and USER.md writes into Ladybug
  • Background prefetch of relevant memories before each turn

When to use it

  • Keep agent memory entirely on your own disk with no hosted provider
  • Link related memories and traverse them as a graph
  • Recover high-importance facts after context compression

Who it is for: Hermes Agent users who want local, inspectable memory with structured recall and no SaaS dependency.

How it fits with Hermes Agent

Built for Hermes Agent as a standalone memory-provider plugin that is configured through hermes memory setup.

How to install Hermes Ladybug Memory

These commands are copied from the project's README. Check the repository for the latest steps before you run them.

hermes plugins install Ladybug-Memory/hermes-memory-plugin
pip install ladybug-memory
ln -s ~/.hermes/plugins/ladybug ~/.hermes/hermes-agent/plugins/memory/ladybug
hermes memory setup

Requirements: Hermes with the memory-provider plugin system and ladybug-memory 0.1.4 or newer.

FAQ

What is Hermes Ladybug Memory?

Hermes Ladybug Memory is a memory-provider plugin that stores Hermes Agent memory in a local Ladybug graph database file, with keyword search, importance scores, typed entries and graph links.

How do I install Hermes Ladybug Memory?

Run hermes plugins install Ladybug-Memory/hermes-memory-plugin and pip install ladybug-memory, link the plugin into the hermes-agent plugins/memory directory, then pick ladybug in hermes memory setup.

Why does Hermes Ladybug Memory need a symlink?

The README says Hermes's memory provider discovery only scans providers bundled in plugins/memory inside the hermes-agent source tree, so the symlink makes the plugin discoverable. It links an upstream feature request, hermes-agent issue 4956.

Similar memory for Hermes Agent

All memory

Related guides: SOUL.md for Hermes Agent: what it is and how to write one · Run multiple Hermes agents with profiles