Hermes Ladybug Memory

Ladybug Memory as a standalone Hermes memory-provider plugin.

It gives Hermes a fully local, file-based memory store backed by LadybugMemory — a columnar embedded graph database (.lbdb). No API keys, no cloud, no sync service. Everything stays on disk in HERMES_HOME.

Why Ladybug Memory

Ladybug Memory is a local-first memory backend for Hermes that keeps all data in a single .lbdb file on your machine. It supports BM25 keyword search, importance-weighted recall, typed memory entries, named graph edges between entries, and optional GLiNER2 entity extraction — all without any external service.

Where It Fits

• Best fit if you want local, inspectable memory with structured recall and graph relationships, and no SaaS dependency.
• Compared with Honcho, Mem0, and RetainDB, this is fully local with no hosted component.
• Compared with Observational Memory and Holographic, this is less about shared cross-agent markdown stores and more about a typed, searchable, graph-linked memory store owned entirely by Hermes.
• Compared with OpenViking and ByteRover, this is simpler and more direct — no hierarchical browser or knowledge-graph UI, just a fast embedded database with the tools the agent needs.

Install

Install the plugin and its Python dependency:

hermes plugins install Ladybug-Memory/hermes-memory-plugin
pip install ladybug-memory

Then link the plugin into the memory provider directory so hermes memory setup can discover it:

ln -s ~/.hermes/plugins/ladybug \
      ~/.hermes/hermes-agent/plugins/memory/ladybug

Why the symlink? Hermes's memory provider system currently only discovers providers bundled in plugins/memory/ inside the hermes-agent source tree. User-installed plugins (~/.hermes/plugins/) are not scanned by the memory discovery system yet. This symlink bridges the gap. See NousResearch/hermes-agent#4956 for the upstream feature request.

Finally, configure it:

hermes memory setup    # select "ladybug"

Requirements

• Hermes with the memory-provider plugin system
• ladybug-memory >= 0.1.4

Install into the Hermes runtime environment if you do not already have it:

pip install ladybug-memory

For GLiNER2 entity extraction (optional):

pip install ladybug-memory[extract]

Manual Install

If you prefer cloning manually:

git clone https://github.com/Ladybug-Memory/hermes-memory-plugin.git \
  ~/.hermes/plugins/ladybug
ln -s ~/.hermes/plugins/ladybug \
      ~/.hermes/hermes-agent/plugins/memory/ladybug
pip install ladybug-memory
hermes memory setup

What It Adds

Tools:

• ladybug_store: persist a new memory entry with type and importance score
• ladybug_search: BM25 keyword search across stored memories
• ladybug_recall: retrieve recent or high-importance memories
• ladybug_update: correct or update a memory by ID
• ladybug_delete: delete a memory by ID
• ladybug_link: create a named relationship between two memories
• ladybug_related: traverse the memory graph by relationship
• ladybug_entity: entity-level KG queries via GLiNER2 (optional)

Memory integration:

• background prefetch before every turn (importance-weighted recall + query search)
• mirrors built-in MEMORY.md / USER.md writes into Ladybug automatically
• surfaces high-importance memories during context compression

Config

All keys go under memory.ladybug in ~/.hermes/config.yaml:

Key	Default	Description
db_path	$HERMES_HOME/ladybug.lbdb	Path to the Ladybug database file
prefetch_limit	6	Memories surfaced before each turn
min_importance	3	Minimum importance score for prefetch recall
auto_link	false	Auto-link mirrored built-in memory writes

Memory Types

general · preference · fact · project · person · event · task

Importance Scores

1–10 scale. Higher scores surface more often in prefetch recall. Built-in MEMORY.md / USER.md mirrors use importance 6 (explicit user signal). Tune with ladybug_update over time.

Validation

This repository ships standalone tests for the provider behavior. Run them with:

uv run --with pytest pytest tests -q

Notes

• This repository is laid out as a Hermes directory plugin, so the repo root is the plugin root.
• The installed plugin name is ladybug, matching the GitHub repo's directory name.
• Hermes currently clones directory plugins from Git but does not install their Python dependencies automatically, so the pip install ladybug-memory step is still required.