MindClaw provides a persistent, structured knowledge layer and memory graph for AI agents to ensure long-term fact retention and conflict detection.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install mindclaw
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install mindclaw using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is MindClaw?
MindClaw is a sophisticated structured long-term knowledge layer designed specifically for AI agents. Unlike standard setups that rely on raw conversational history, MindClaw stores curated facts, decisions, and complex relationships. It enhances Openclaw Skills by providing a dedicated memory graph equipped with metadata features such as conflict detection, importance scoring, and confirmation reinforcement.
By synchronizing data back to the primary MEMORY.md file, MindClaw ensures that all structured intelligence is searchable via native agent tools. It acts as the brain for your agents, allowing them to remember critical project details while intelligently managing information decay and relevance through advanced algorithms like MMR diversity and temporal decay.
MindClaw Use Cases
- Persistent storage of user preferences and project-specific decisions across multiple agent sessions.
- Automated extraction of structured facts from unstructured conversation text to build a reliable knowledge base.
- Building a relational knowledge graph where Openclaw Skills can link disparate pieces of information for better context.
- Managing high-density information environments where temporal decay is needed to prioritize recent or important facts.
- Cross-tool visibility by syncing structured memories into standard Markdown formats for other agent tools to consume.
How MindClaw Works
- Context Retrieval: The agent utilizes the context_block tool to fetch a token-optimized block of relevant memory before responding to a prompt.
- Knowledge Capture: During or after an interaction, the agent uses the capture or remember tools to save key facts, decisions, or preferences.
- Metadata Assignment: Every memory is assigned metadata, including importance scores and timestamps, to support sophisticated recall logic.
- Verification: The agent can use the confirm tool to reinforce memories that have proven correct, increasing their longevity and search priority.
- Synchronization: All structured data is synced to the OpenClaw ecosystem via the sync_openclaw command, making memories available in the local MEMORY.md file.
- Maintenance: Periodic consolidation runs automatically to deduplicate entries and link related concepts within the knowledge graph.
MindClaw Setup
To integrate MindClaw into your Openclaw Skills workflow, install the package and run the interactive setup wizard:
pip install mindclaw[mcp] && mindclaw setup
The setup utility configures your workspace paths, agent identification, and registers the MCP server with your environment. For enhanced semantic search, ensure Ollama is running locally as MindClaw will auto-detect it for zero-dependency embedding generation.
MindClaw Data Schema & Taxonomy
MindClaw organizes its data using a structured approach that prioritizes both local searchability and metadata richness:
| Component | Description |
|---|---|
| SQLite DB | Per-agent isolated storage for structured facts and metadata. |
| Knowledge Graph | A relational map connecting related memories via the link tool. |
| MEMORY.md | A Markdown-based mirror for compatibility with standard Openclaw Skills. |
| config.json | Centralized configuration stored in ~/.mindclaw/ for workspace and agent settings. |
Data includes importance scores, temporal decay values, and semantic embeddings for hybrid search capabilities.
MindClaw Advanced Features
- Hybrid Search Engine: Combines BM25 keyword search with semantic embeddings and MMR diversity for highly relevant memory recall.
- Memory Pinning: Allows users to mark specific facts as permanent, exempting them from temporal decay or archival.
- Timeline Reconstruction: Enables agents to reconstruct events and decisions chronologically over specific time windows.
- Automatic Consolidation: Uses intelligent deduplication to merge near-duplicate memories and maintain a clean knowledge graph.
- Conflict Detection: Identifies and flags potential contradictions when new memories are captured against existing facts.
SKILL.md
Loading
Related Openclaw Skills
MiniMax Music Maker
A powerful AI music generation skill that transforms lyrics and style prompts into high-quality audio tracks using the MiniMax music-2.5 model.
MiniMax Music Expert
An advanced AI music generation skill that transforms text prompts and lyrics into professional-grade audio tracks using MiniMax models.
mm-easy-voice
A high-performance text-to-speech tool for Openclaw Skills that converts text to natural audio using the MiniMax Voice API.
beepctl
A command-line interface for the Beeper Desktop API that enables unified messaging across platforms like Telegram, WhatsApp, and Slack.
xAI Grok Integration
A powerful interface to interact with xAI Grok models and perform real-time searches across X/Twitter posts.
Search X
A powerful search integration that enables AI agents to query X/Twitter in real-time using either xAI Grok or the native X API.
Featured*
