Meditate for Openclaw

A proactive reflection skill that enables AI agents to generate sandboxed insights and strategic questions during idle time.

ivangdavila
v1.0.1
Feb 18, 2026
2
1.1k
0

Install & Download

1. ClawHub CLI

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

npx clawhub@latest install meditate

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 meditate 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 Meditate?

Meditate is a specialized tool designed for Openclaw Skills that transforms an AI agent from a reactive assistant into a proactive thinking partner. By analyzing conversation patterns and user history during idle periods, the skill generates thoughtful observations and open-ended questions stored in a dedicated local sandbox. This process allows the agent to identify architectural gaps, strategy improvements, or habit patterns without ever executing unrequested actions.

The skill operates under a strict reflection-only policy, ensuring that all outputs remain non-executable and purely discursive. It utilizes an adaptive rhythm that adjusts the frequency of meditations based on user engagement levels and feedback, making it a high-value addition to any Openclaw Skills library for users seeking deeper insights from their AI interactions.

Meditate Use Cases

  • Identifying technical debt or architectural inconsistencies in developer workflows.
  • Surfacing strategic gaps for entrepreneurs based on recent project discussions.
  • Analyzing prompt patterns and style evolution for creative professionals.
  • Summarizing habit trends or goal progress from personal interactions.
  • Maintaining a queue of high-value insights to be presented during the next active session.

How Meditate Works

  1. The agent monitors idle time between user interactions to trigger the meditation cycle.
  2. It analyzes conversation history and local memory files to identify relevant patterns or recurring topics.
  3. A sandboxed reflection is generated as pure text, focusing on observations and questions rather than code or actions.
  4. The insight is stored in a local queue within the ~/meditate/ directory, respecting a maximum limit of three pending items.
  5. Upon the next user interaction, the agent presents the oldest insight and waits for feedback to refine future reflections.

Meditate Setup

To get started, create the necessary directory structure and configuration files in your home directory:

mkdir -p ~/meditate/archive
touch ~/meditate/profile.md ~/meditate/topics.md ~/meditate/insights.md ~/meditate/feedback.md

Ensure your Openclaw Skills environment has read/write access to the ~/meditate/ path and initialize your profile.md with your preferred focus areas.

Meditate Data Schema & Taxonomy

File Purpose Key Metadata
profile.md User identity & preferences User type, focus areas, rhythm
topics.md Meditation backlog Active topics, priority levels
insights.md Delivery queue Pending observations, context
feedback.md Quality control User reactions, rhythm adjustments
archive/ Historical logs Past insights (30-day retention)

Meditate Advanced Features

  • Adaptive Rhythm: Automatically scales meditation frequency (from nightly to weekly) based on interaction density and user feedback.
  • Profile Detection: Dynamically categorizes focus areas into System, Personal, Creative, Developer, or Entrepreneur profiles based on conversation context.
  • Feedback-Driven Prioritization: Demotes or removes topics automatically when the user indicates a lack of relevance.
  • Absolute Sandboxing: A core rule-set that prevents the generation of commands, scripts, or actionable code, ensuring the skill remains purely reflective.

SKILL.md


Loading

Related Openclaw Skills

METADATA

Github Stars: 0
forks: 0

Featured*