Agent skill · software engineering · aradotso
mempalace-ai-memory
mempalace-ai-memory — an agent skill for coding agents.
Why this skill is useful
Adds executable scripts and commands for managing a local AI memory system that enhances conversation context retrieval and storage capabilities.
What it needs
Requires mempalace installed locally. About 6k tokens when loaded. Last updated 2026-07-09. 70 stars on the source repository.
What this skill does
bash pip install mempalace bash mempalace --version mempalace status bash Initialize for a specific project directory mempalace init ~/projects/myapp This creates ~/.mempalace/palace/ with the wing/room/closet/drawer structure MemPalace auto-detects rooms (topics) from the project bash Mode: projects — code, docs, notes mempalace mine ~/projects/myapp Mode: convos — Claude, ChatGPT, Slack exports mempalace mine ~/chats/ --mode convos Mode: general — auto-classifies into decisions, preferences, milestones, problems, and emotional context mempalace mine ~/chats/ --mode convos --extract general bash mempalace status Shows wings, rooms, and memory counts bash Add MemPalace as an MCP server in Claude Code claude mcp add mempalace -- python -m mempalace.mcpserver python from mempalace.searcher import searchmemories results = searchmemories( query="why did we switch to GraphQL", palacepath="~/.mempalace/palace", topk=5 ) for result in results: print(result["content"]) print(result["source"]) # drawer path print(result["score"]) # relevance score python from mempalace.memory import storememory storememory( content="Decided to use PostgreSQL over MySQL for JSONB support.", wing="myapp-project", room="decisions", palacepath="~/.mempalace/palace" ) python from mempalace.wakeup import generatewakeup context = generatewakeup( palacepath="~/.mempalace/palace", useaaak=True # AAAK compression — ~170 tokens ) Inject into your local LLM's system prompt systemprompt = f"You have access to the following memory context:\n{context}" python from mempalace.miner import minedirectory minedirectory( sourcepath="~/chats/", mode="convos", extract="general", # decisions, preferences, milestones, problems palacepath="~/.mempalace/palace" ) bash mempalace wake-up > context.txt python import subprocess context = subprocess.checkoutput(["mempalace", "wake-up"]).decode() Inject into Ollama / llama.cpp / any local LLM import ollama response = ollama.chat( model="llama3", messages=[ {"role": "system" …
How to use it
Reference it in AdaL, Claude Code, Cursor or any coding agent — nothing to install:
@skills aradotso/mempalace-ai-memory