mema

What This Skill Does

Mema is OpenClaw's centralized memory system, acting as a personal brain for your agent. It integrates two distinct storage backends to ensure your workspace remains organized and your session context remains persistent across restarts. At its core, Mema provides a SQLite-backed Metadata Index that keeps track of the file paths, titles, and descriptive tags of your important documents. It is important to note that this index is strictly for metadata; it does not store the full content of your files, making it a high-performance solution for document retrieval without the bloat of indexing large file contents.

Beyond file organization, Mema leverages Redis as a short-term memory buffer. This allows the OpenClaw agent to maintain an ephemeral mental state, such as current task progress or specific user preferences, using a key-value store with a default 6-hour TTL. By separating durable metadata from transient state, Mema keeps your agent's knowledge environment tidy and performant.

Installation

1. Install via the package manager: clawhub install openclaw/skills/skills/1999azzar/mema
2. Copy the configuration file: cp env.example.txt .env
3. Configure your Redis environment variables in the .env file (ensure REDIS_HOST and REDIS_PORT are set).
4. Initialize the SQLite memory database: python3 $WORKSPACE/skills/mema/scripts/mema.py init

Use Cases

• Project Knowledge Management: Quickly catalog research papers, design docs, or code snippets across scattered project directories.
• Agent Context Persistence: Store session-specific variables or temporary status flags to ensure the agent picks up exactly where it left off after an inactivity period.
• Tag-based Retrieval: Organize disparate files into logical groups (e.g., 'research', 'todo', 'urgent') and retrieve lists of paths matching those tags.

Example Prompts

1. "Mema, index the file at ./docs/project_spec.md with the tag 'architecture'."
2. "List all files in my memory bank tagged with 'research' so I can review them."
3. "Set the current session status in Mema to 'Drafting Phase' with a 2-hour TTL."

Tips & Limitations

• Performance: Because Mema uses SQLite, it is extremely fast for metadata lookups. Keep your tags descriptive but concise to improve filter performance.
• Security: Mema operates entirely within local directories (~/.openclaw/memory), ensuring your metadata does not leak to external services. Use trusted Redis instances only.
• Data Scope: Mema is an indexing tool. If you need to search the actual text content inside your files, you should use a dedicated vector search or full-text indexing tool alongside Mema. Always remember that Redis entries are ephemeral; do not store permanent configuration data in the mema:mental namespace.