Ladybug Memory adds an independent, graph-native memory layer to Agent Zero using LadybugDB.
- It runs in parallel with built-in
_memoryinstead of replacing it. - If
_memoryis disabled, this plugin can still memorize and recall. - It stores memories as graph nodes connected to topic nodes.
- It includes a dashboard with:
- memory search/list view
- graph visualization of memory-topic relations
helpers/ladybug_memory.py: storage + retrieval + graph projection logic.extensions/python/...: runtime integration (init, memorize, recall).api/ladybug_memory_dashboard.py: dashboard API actions.webui/: dashboard + settings.
The plugin uses the Python package real_ladybug.
- Install is automatic through
hooks.py -> install()when the plugin is installed or updated. - Uninstall runs
hooks.py -> uninstall()and performs cleanup for project workspaces. - If automatic install fails, install manually in the framework runtime:
python -m pip install "real_ladybug>=0.14.1"
- Database files are stored under plugin-owned paths:
- global:
usr/plugins/ladybug_memory/data/<subdir>/ladybug_memory.lbug - project-isolated:
<project>/.a0proj/ladybug_memory/ladybug_memory.lbug
- global:
- No files are written outside plugin-owned or project-owned metadata paths.