Use this skill to set up a memory layout that is easy to retrieve from, audit, and compress.
When the user asks to apply this skill (for example: Apply dr-memory-foundation to this workspace), do this:
1) Inspect the existing workspace memory files.
2) Create any missing template files from references/templates/.
3) Preserve existing notes; merge or relocate content rather than deleting it.
4) Normalize MEMORY.md into preferences + indexes only.
5) Ensure memory/always_on.md contains a tiny policy header + topic catalog.
6) Confirm what changed.
This apply flow should be idempotent: if the structure already exists, do not duplicate sections or overwrite user content blindly.
Run this from the workspace root when you want the templates created automatically:
cd ~/.openclaw/workspace
python3 ./skills/dr-memory-foundation/scripts/install_memory_foundation.py
tree -L 2 memory | head -n 40
If files already exist the script skips them; otherwise it copies the templates from references/templates/.
MEMORY.md (small): preferences + indexes only.memory/always_on.md: tiny policy header + topic catalog (with keywords).memory/now.mdmemory/open-loops.mdmemory/automation.mdmemory/topics/glossary.mdmemory/topics/.md memory/YYYY-MM-DD.md1) Create the folders/files from references/templates/.
2) Move existing knowledge into topic files without deleting source logs.
3) Keep MEMORY.md as indexes + preferences only.
4) Update the topic catalog in memory/always_on.md as topics evolve.
Use the files under references/templates/.
共 3 个版本