CompoundMind for Openclaw

An experience distillation engine that turns raw daily memory logs into compounding intelligence for AI agents.

cassh100k
v0.1.0
Feb 26, 2026
0
1.3k
0

Install & Download

1. ClawHub CLI

The fastest way to install a skill directly from the registry.

npx clawhub@latest install compound-mind

2. Manual Installation

Copy the skill folder to one of these locations

Global
~/.openclaw/skills/
Workspace
<project>/skills/

Priority: Workspace > Local > Bundled

3. Prompt Installation

Copy this prompt to OpenClaw to install it automatically.

Help me install compound-mind 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 CompoundMind?

CompoundMind is a specialized engine designed to overcome the limitations of session-based memory by distilling raw interaction logs into a structured, searchable intelligence layer. As a core component of the Openclaw Skills library, it extracts lessons, decisions, and skill updates to ensure that agents grow smarter over time rather than repeating mistakes. By processing logs through Claude Haiku, it creates a high-density wisdom index that allows agents to learn from their own history.

CompoundMind Use Cases

  • Surfacing past configuration patterns and API keys during complex technical tasks.
  • Briefing an agent on historical project context before starting a new coding session.
  • Analyzing agent performance to identify and rectify recurring logic errors or failed patterns.
  • Maintaining a persistent record of interpersonal preferences and communication styles for long-term collaborations.

How CompoundMind Works

  1. Raw memory files are parsed and distilled into categorized wisdom blocks (lessons, facts, decisions) using an LLM-based extraction pipeline.
  2. New intelligence is synchronized with a local SQLite database utilizing FTS5 for high-performance searching across your Openclaw Skills workspace.
  3. Task-specific briefings are generated by querying the wisdom index for the most relevant past experiences based on the current prompt.
  4. Longitudinal data is analyzed to produce growth reports and identify recurring mistake patterns through the growth tracker module.

CompoundMind Setup

Ensure you are operating within your agent environment and have Python 3.10+ installed with the Anthropic SDK configured.

# Navigate to the skill workspace
cd /root/.openclaw/workspace/compound-mind

# Run the full sync pipeline to process existing memory files and build the index
python compound_mind.py sync

# Generate a pre-task briefing to load relevant context
python compound_mind.py brief "debug a Python cron job"

CompoundMind Data Schema & Taxonomy

The skill organizes its intelligence into a structured local directory, ensuring data privacy and low-latency access without external vector databases.

Component Location Format Purpose
Wisdom Index data/wisdom.db SQLite FTS5 Full-text searchable database of all distilled experiences.
Experience Logs data/experiences/ JSON Per-source distilled outputs containing structured lessons.
Growth State data/growth.json JSON Persistent tracking of capability metrics and mistake patterns.
Briefing Archive data/briefs/ Markdown Saved pre-session briefings for historical reference.

CompoundMind Advanced Features

  • Intelligent recency-weighted search scoring using BM25 and exponential decay algorithms.
  • Automated mistake detection that flags negative lesson patterns occurring across multiple dates.
  • Incremental processing via file-hash tracking to ensure only new or modified logs are distilled.
  • Seamless integration for Openclaw Skills users to maintain persistent capability scores (0-100) based on actual decision outcomes.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*