A meta-skill for AI agents designed to expand functional capabilities through skill discovery and ensure continuous self-improvement by capturing durable lessons.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-essentials
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 agent-essentials 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 Agent Essentials?
Agent Essentials serves as a foundational meta-layer for AI coding agents, enabling them to overcome native limitations by proactively searching for and integrating new capabilities. Instead of failing when a request falls outside its built-in scope, an agent equipped with this skill identifies the gap and searches for relevant Openclaw Skills to bridge it.
Beyond expansion, it facilitates a robust self-improvement loop. By monitoring for meaningful failures or user corrections, it distills complex experiences into concise, actionable lessons. These insights are then routed to durable documentation like AGENTS.md or SOUL.md, ensuring the agent evolves alongside the user's specific workflows and preferences.
Agent Essentials Use Cases
- Automating complex tasks that require external platform integrations or specific third-party tools.
- Identifying and filling capability gaps by searching for existing Openclaw Skills on ClawHub.
- Capturing recurring developer preferences to ensure consistent coding style and project-specific logic.
- Reducing technical debt by documenting 'gotchas' and workflow rules in durable markdown files.
- Transforming repetitive user corrections into permanent behavioral changes for the AI agent.
How Agent Essentials Works
- Gap Detection: The agent analyzes user requests for keywords like 'automate' or 'integrate' to determine if a missing capability or repeatable process exists.
- Skill Discovery: If a gap is identified, the agent searches local libraries and ClawHub for matching Openclaw Skills that can resolve the task.
- User Confirmation: The agent presents potential candidates and waits for explicit user approval before installing a new skill or scaffolding a custom one.
- Lesson Capture: During the interaction, the agent monitors for meaningful failures or corrections, distilling them into a 3-line 'What/Correct/Next time' format.
- Durable Routing: Validated lessons are promoted to specific files like AGENTS.md, TOOLS.md, or USER.md to ensure long-term behavioral persistence.
Agent Essentials Setup
To enable the self-improvement and discovery features of this skill, ensure your agent has access to your local configuration directory. No complex installation is required; simply include the skill in your agent's active profile.
# Ensure your local durable files exist or can be created
touch AGENTS.md TOOLS.md SOUL.md
# The agent will automatically check these paths:
# 1. ./
# 2. ~/.claude/
Agent Essentials Data Schema & Taxonomy
Agent Essentials organizes information across several durable markdown files to maintain a clear taxonomy of knowledge:
| File Name | Purpose | Scope |
|---|---|---|
AGENTS.md |
General workflow rules and agent instructions | Project/Local |
TOOLS.md |
Specific technical 'gotchas' and tool-related caveats | Project/Local |
SOUL.md |
Persona, voice, and boundary patterns | Project/Local |
USER.md |
Persistent user preferences and long-term memory | User-scoped |
YYYY-MM-DD.md |
Daily session notes and transient learnings | Temporal |
Agent Essentials Advanced Features
- ClawHub Integration: Native ability to query the central repository for the latest Openclaw Skills based on real-time task requirements.
- Automated Lesson Promotion: Intelligent filtering that only promotes high-value, recurring lessons to durable files to prevent documentation bloat.
- Project-Root Scoping: Automatically detects if it should apply rules globally at the user level or locally within a specific project repository.
- Multi-Agent Coordination: Provides a standardized way for different agent instances to share learnings through the AGENTS.md protocol.
SKILL.md
Loading
Related Openclaw Skills
Novel Writer
A creative writing specialist designed to craft high-pacing urban science fiction novels with a blend of daily life and futuristic elements.
Custom Grok Search
A powerful search utility enabling AI agents to access real-time web data and X/Twitter discussions via xAI Grok APIs.
Smart Parenting Feynman
A sophisticated AI persona skill that combines the Feynman technique with a warm, playful Chinese parenting style for teaching and creative orchestration.
ClawKB
ClawKB enables AI agents to interact with a centralized knowledge base for automated registration, research capture, and data retrieval via HTTP.
SimpleTexting
A robust integration for SimpleTexting that enables AI agents to manage SMS marketing campaigns, contact lists, and automated messaging.
pdf2md
A high-performance PDF to Markdown converter that extracts text, tables, and complex formulas using the MinerU Open API.
Featured*
